Drones with obstacle detection and collision avoidance sensors are becoming more prevalent in both the consumer and professional sectors. This year, we have quite a few drones with collision avoidance technology.
This obstacle detection and avoidance technology started with sensors detecting objects in front of the drone.
Now the latest drones from DJI, Walkera, Yuneec and others have front, back, below and side obstacle avoidance sensors.
At the time of writing there is only 2 drones, which has all 6 directions of obstacle detection.
In this up to date article, we take a quick view at the top drones with obstacle detection and collision avoidance technology. We also give you a brief overview of the type of obstacle detection sensors being used, including information on software algorithms and SLAM technology, which is used to interpret the images being scanned by the sensors.
There are also links and information if you would like to create your own DIY collision avoidance system.
Surprisingly, there is not just one type of obstacle detection sensor being used by the drone manufacturers.
We are seeing Stereo Vision, Monocular Vision, Ultrasonic, Infrared, Time-of-Flight and Lidar sensors being used to detect and avoid obstacles. Manufacturers are fusing these various sensors together to create the obstacle detection and collision avoidance systems.
12 Top Drones With Obstacle Avoidance
The below obstacle avoidance drones contain from 1 to 6 directions of obstacle avoidance technology. We will review this list in more detail further along this post.
- Skydio 2 (New)
- Kespry 2.
- DJI Mavic Air.
- Walkera Vitus.
- DJI Mavic Pro.
- DJI Mavic 2 Pro & Zoom.
- Yuneec Typhoon H / H Plus.
- DJI Phantom 4 Pro.
- Walkera Voyager 5.
- DJI Matrice 200.
- DJI Inspire 2.
- Autel Evo.
As you can see DJI who are the leading consumer and professional drone manufacturer with something like 70% of the market are also leading the way when it comes to obstacle avoidance drones.
The Skydio 2 was only released in October 2019. It has overlapping visual sensing meaning it has full obstacle detection. It uses a super fisheye lens for 360° vision for true omnidirectional vision including above and below obstacle detection. The Skydio 2 was particularly designed for tracking people and subjects.
The Mavic 2 have obstacle detection all 6 sides of the drone. The Mavic 2 can also fly around obstacles in front or behind it. We have more information on the Mavic 2 obstacle sensing system further down. The DJI Mavic Pro 2 and Mavic 2 Zoom released in August 2018 have new cameras and superb stabilization. Click on the link above to watch videos on these latest DJI Mavic 2 quadcopters.
Next best is the DJI Phantom 4 Pro obstacle avoidance system. It has 5 directions of obstacle sensing and 4 directions of obstacle avoidance, which is outstanding. It also has many intelligent flight modes, super smooth stability and top 4k camera.
Each drone is different with the more expensive drones been used for commercial inspections, photogrammetry and movie making.
One of the best obstacle detection and collision avoidance on a small drone is the DJI Mavic Air and Mavic Air 2 which was released in April 2020.
Both Mavic Air models have 3 directions of collision avoidance, which are forward, backward and downwards sensing. It’s sense and avoid system was the first to actually of the small drones to detect obstacles and then to fly around the object.
The Mavic Air has some terrific innovative technology and is a terrific small sized drone. You can read a full Mavic Air review here and the new Mavic Air 2 is even more impressive. You can read the full Mavic Air 2 review here, which has some terrific videos.
Below, I explain the Mavic Air obstacle detection and collision avoidance technology in more detail, along with the other drones above.
Benefits Of Collision Avoidance Drones
There are so many advantages and benefits to drones with collision avoidance systems.
Safer Drones With Obstacle Detection
Less drone crashes is what everyone wants. For the drone owner, it is very easy to get carried away while flying. If you loose your bearings or concentration, you could easily fly backwards or sideways into an object. It is even possible to fly head on into an obstacle especially when flying further out distances.
Nearly all drones have first person view which transmit the video from the drone camera back to the remote controller, smartphone or tablet. However, it is possible to loose this video transmission.
If you have flown a good bit out of direct line of sight, without obstacle avoidance it is going to be impossible to fly back home safely without video transmission. Pressing the Return To Home button is the only option but if you don’t have obstacle avoidance it could very well crash.
Drones are being used in many public areas and at events as they capture stunning film from unique angles. Unfortunately, there have been a few accidents which is not good. People should be safe at concerts or sporting events so collision avoidance drones at these events is a must.
Most drones today fly using the GPS and GLONASS satellite navigation systems to know exactly where it is and to fly stable. Flying outdoors in open space is easy. The big challenge is flying indoors. There are many great uses for drones to fly indoors.
We are seeing factories and warehouses looking to use drones in many ways such as inspections, counting inventory and logistics.
Flying indoors is more difficult. Less space and more obstacles are the biggest problems. Many drones need pilots to manually fly indoors. With obstacle avoidance sensors, this will allow drones to navigate autonomously indoors.
The costs for insuring a professional aerial filming or multispectral drone can be quite high. A top of the range multirotor carrying expensive camera equipment could cost up anywhere up to USD 50k. For these drones, it is essential to have insurance and the insurance costs are high. Having a drone with obstacle detection collision avoidance systems will bring down these insurance costs.
Obstacle Detection Drones For Ease Of Mind
The latest top drones today have 4k cameras and film beautifully. Many people would love to own a drone but are terrified of crashing. If a drone pilot crashes into a tree, then it is pretty bad, but if it crashes into a person, cyclist or car it could be pretty catastrophic as well as highly embarrassing. Many people are afraid that they will crash on their first flight wiping out their purchase.
With obstacle detection along with the many safety features which come on drones today, we should see many more people taking up drone flying as a hobby or as a profession. There as so many great uses for drones and even more to be realized.
Future – Safe Autonomous Drones Delivering Parcels
Drones are with us to stay and we could be looking at a future where drones are autonomously delivering parcels, medications and pizza to our doors. There are so many challenges to be overcome for this to happen. Without doubt, drones will need to be 100% safe. They will need to be perfect at avoiding obstacles both moving and stationery.
Obstacle Avoidance Sensors
The various drones are using the following obstacle avoidance sensors either on their own or combined;
- Stereo Vision.
- Ultrasonic (Sonar).
- Monocular Vision.
There are brief easy to understand explanations on how each sensor works further down this post.
What Is Obstacle Detection And Collision Avoidance Technology
For a drone, car or robot to detect objects and then take action to avoid the obstacle whether to stop, go around or above the object involves many complex technologies working together to create an integrated system. This entails many various sensors, software programming which include mathematical modelling, algorithms, machine learning and aspects of SLAM technology. Let’s take a quick look at these various technologies.
Sensor fusion is a process by which data from several different sensors are “fused” to compute something more than could be determined by any one sensor alone. Sensor fusion is a subcategory of data fusion and is also called multisensory data fusion or sensor-data fusion. Many of the DJI drones combine various sensors into their collision avoidance system.
Another area where sensor fusion is used is in precision agricultural using multispectral sensors on drones. Multispectral remote sensing imaging technology use Green, Red, Red-Edge and Near Infrared wavebands to capture both visible and invisible images of crops and vegetation.
These various obstacle avoidance sensors feed the data back to the flight controller which is running obstacle detection software and algorithms. The flight controller has many functions. One of these is to process image data of surroundings which was scanned by the obstacle detection sensors in realtime.
Obstacle Avoidance Algorithms
The obstacle avoidance algorithm is the process or set of rules to be followed in calculating the data from the various sensors. The algorithm is a detailed step-by-step instruction set or formula for solving the problem of detecting all types of objects both moving or stationary.
Depending on the algorithm, it will be able to compare real time data from stored referenced images of objects and can even build on these images.
There are many techniques which can be used for obstacle avoidance including how the algorithm processes the data. The best technique depend on the specific environment and is different for a collision avoidance drone and a robot in a factory.
Here is a nice web page, which explains obstacle avoidance techniques. It gives you an idea of the technology and techniques which is used to detect objects in a very simple way.
The algorithm is very important. You could have the best obstacle detection sensor but if the software and algorithm is poorly written, then the data from the sensor will not be interpreted incorrectly leading to flight errors and the drone crashing.
SLAM Technology For Detecting And Avoiding Obstacles
Simultaneous localization and mapping or SLAM is an extremely important technology when it comes to drones, cars and robots in detecting and avoiding obstacles.
SLAM is a process whereby a robot or a device can create a map of its surroundings, and orient itself properly within this map in real time. This is no easy task, and SLAM is currently at the forefront of technology research and design.
SLAM technology works by first building a pre-existing map of its environment. The device such as a drone or robot is programmed with pre-existing maps. This map is then refined as the robot or drone moves through the environment.
The true challenge of this technology is one of accuracy. Measurements must constantly be taken as the robot or drone as it moves through its space, and the technology must take into account the “noise” that is introduced by both the movement of the device and the inaccuracy of the measurement method.
SLAM is fascinating technology and you can read more about it in this article entitled “What is SLAM Technology“. Many of the obstacle detection and avoidance technology in drones use some parts of SLAM. Monocular vision is one such technology.
Full Obstacle Avoidance System – Flight Controller
Each drone will have slight differences on what to do once an object has been detected. The sensors scan the surroundings and feed this information back to the flight control system which will control the obstacle avoidance algorithm. The flight controller will then direct the drone depending on interpretation of the visual data from the algorithm whether to fly around, above or just hover in front of the obstacle.
Obstacle Detection To Track And Follow Objects
These obstacle detection sensors can do more than just detect objects and navigate around them or to stop from crashing into the obstacle. All of the drones listed above use their vision sensors together with advanced image recognition algorithms to allow the quadcopter to recognize and tracks objects. These obstacle detection sensors and algorithms can detect people, vehicles, animals and many other objects to follow.
On the DJI drones, this technology is known as ActiveTrack with the following choices
Trace – Follow behind or in front of a subject, avoiding obstacles automatically.
Profile – Fly alongside a subject at a variety of angles to get profile shots of the subject.
Spotlight – Keep the camera trained on a subject while the aircraft flies almost anywhere.
Ultrasonic sensors beneath the Phantom 4 and Mavic allow these drones to track the level of ground with a terrain follow mode. Basically these drones stay at the same height level above the ground automatically.
How Collision Avoidance Sensors Work
Next, we give a brief explanation how each obstacle detection sensor works. We have links to further articles and videos relating to Stereo Vision, Infrared, Lidar, ToF, Ultrasonic and Monocular vision sensors.
Stereo Vision Sensors For Obstacle Avoidance
Stereo vision works in a similar way to 3D sensing in our human vision. Stereoscopic vision is the calculation of depth information by combining two-dimensional images from two cameras at slightly different viewpoints.
It begins with identifying image pixels which correspond to the same point in a physical scene observed by multiple cameras. The 3D position of a point can then be established by triangulation using a ray from each camera.
The more corresponding pixels identified, the more 3D points which can be determined with a single set of images. Correlation stereo methods attempt to obtain correspondences for every pixel in the stereo image, resulting in tens of thousands of 3D values generated with every stereo image.
DJI use stereo vision for obstacle avoidance on the front of their drones. They also combine Stereo Vision and Ultrasonic sensors beneath their drones too.
Here is a short video on how Stereo Vision works.
Centeye RockCreek Vision Sensor
Centeye has prototyped a vision based system to allow small drones to both hover in place without GPS and avoid collisions with nearby obstacles.
This system was tested on a “nano” unmanned aerial vehicles (UAV) which weighs about an ounce and can fit in the palm of one’s hand. It uses Centeye RockCreek™ vision chips.
The video below shows sample flights in an indoor residence.
Ultrasonic Sensors For Detecting Objects (Sonar)
An Ultrasonic Sensor sends out a high-frequency sound pulse and then times how long it takes for the echo of the sound to reflect back. The ultrasound sensor has 2 openings. One of these openings transmits the ultrasonic waves, (like a tiny speaker) and the other opening receives the ultrasonic waves, (like a tiny microphone).
The speed of sound is approximately 341 meters (1100 feet) per second in air. The ultrasonic sensor uses this information along with the time difference between sending and receiving the sound pulse to determine the distance to an object. It uses the following mathematical equation:
Distance = Time x Speed of Sound / by 2
- Time = the time between when an ultrasonic wave is transmitted and when it is received
- You divide this number by 2 because the sound wave has to travel to the object and back
Most drones use the ultrasonic sensors on the bottom of the drone for detecting ground and also for use in terrain follow mode.
Ultrasound is used in many different fields. Ultrasonic devices are used to detect objects and measure distances. Ultrasound imaging or sonography is often used in medicine. In the nondestructive testing of products and structures, ultrasound is used to detect invisible flaws.
Ultrasound has many industrial uses from cleaning, mixing, and to accelerate chemical processes. Animals such as bats and porpoises use ultrasound for locating prey and obstacles.
The term sonar is used for the equipment used to generate and receive the sound. The acoustic frequencies used in sonar systems vary from very low infrasonic to extremely high ultrasonic.
HC-SR04 Ultrasonic Sensor
The HC-SR04 ultrasonic sensor uses sonar to determine the distance to an object like the way bats do. It offers excellent non-contact range detection with high accuracy and stable readings in an easy-to-use package. From 2 cm to 400 cm or 1 inch to 13 feet.
This HC-SR04 operation is not affected by sunlight or black material like Sharp rangefinders are (although acoustically soft materials like cloth can be difficult to detect). It comes complete with ultrasonic transmitter and receiver module.
You can read a complete guide to the HC-SR04 ultrasonic sensor here.
Time-of-Flight (ToF) Sensors For Collision Avoidance
A Time-of-Flight camera consists of a lens, an integrated light source, sensor and an interface. It is able to capture depth and intensity information simultaneously for every pixel in the image making it extremely fast with high frame rates.
ToF sensors capture depth independently allowing for relatively simply obstacle avoidance algorithms to be used. ToF cameras are also highly accurate.
Time of Flight are also referred to as “Flash Lidar” but this technology should not be confused with Lidar which I discuss further down.
How it works. The ToF camera illuminates the whole scene including objects using a pulse or continuous wave light source and then observing the reflected light.
It measures the time of flight of the pulse from the emitter to the object and then back after reflecting off the object. Because the speed of light is known, the distance to all of the points on the obstacle can be easily calculated.
From these calculations the result is a 3D depth range map which was created in a single shot of an area or scene. It is the quickest technology to capture 3D information.
The Walkera Vitus is using ToF sensors for collision avoidance on the front, left and right hand side of their latest pocket sized quadcopter.
You can read further on the many terrific uses for Flash Lidar ToF sensors on drones here.
AMS ToF Obstacle Detection Sensors
The AMS ToF sensor for obstacle detection and collision avoidance proprietary SPAD (Single Photon Avalanche Photodiode) pixel design and extremely narrow pulse width. measure the direct of VCSEL emitter’s ray reflected from an object.
This low-power time-of-flight sensing technology from AMS enables host systems to measure distances accurately and at very high speed. Accurate distance measurements are used in various applications including presence detection, user face recognition and advanced cameras.
AMS sensors use sophisticated histogram data and smart software algorithms in its ToF sensors giving the following features:
- are able to detect and cancel out the effect of cover glass.
- are immune to smudges and to crosstalk caused by reflections from the cover glass.
- accommodate a large air gap.
- maintain accurate distance detection independent of the object’s color, reflectivity and texture.
- can measure the distance from multiple objects in the field of view.
Infrared Sensor For Obstacle Detection
An Infrared (IR) obstacle detection sensor works in accordance with the infrared reflection principle to detect obstacles.
An IR obstacle avoidance sensor mainly consists of an infrared transmitter, an infrared receiver and a potentiometer. According to the reflecting character of an object, if there is no obstacle, the emitted infrared ray will weaken with the distance it spreads and finally disappear.
If there is an obstacle, when the infrared ray encounters it, the ray will be reflected back to the infrared receiver. Then the infrared receiver detects this signal and confirms an obstacle in front.
To prevent the IR sensor from being confused by visible light, infrared detectors work with a specific frequency of infrared which is produced by the emitter, reflected by an object, then picked up by the receiver. The two devices (emitter and receiver) are matched for optimal sensitivity.
When there is no object, the infrared receiver receives no signals. When there is an object ahead which will block the IR light and then reflect the infrared light back to the receiver.
Here is a nice short video which explains how an IR obstacle sensor works.
Sharp GP2Y0A02YK0F Infrared Distance Sensor
The Sharp GP2Y0A02YK0F measures distances in the 6 to 60 inch (20 – 150 cm) range using a reflected beam of infrared light. By using triangulation to calculate the distance measured, this sensor can provide consistent readings which are less influenced by surface reflectivity, operating time, or environmental temperature.
The Sharp GP2Y0A02YK0F outputs an analog voltage corresponding to the distance to the reflecting object. You can read more on this Sharp IR distance sensor here.
Arduino Nano Board And IR Obstacle Avoidance Sensor Module
Lidar For Obstacle Detection
A lidar sensor calculates distances and detects objects by measuring the time which it takes for a short laser pulse to travel from the sensor to an object and back, calculating the distance from the known speed of light.
Top of the range sensors such as the Velodyne Lidar sensor used in the Google driver-less cars combine multiple laser/detector pairs (up to 64) into one sensor and each can pulse at 20 kHz. This allows for measurements of up to 1.3 million data points per second.
Different applications require different demands on the data quality. However an abundance of data is absolutely necessary for the most reliable object detection making lidar sensors perfect for obstacle detection.
Lidar sensors on drones has many uses and you can read further on lidar sensors on drones here.
The Kespry 2.0 commercial UAV uses Lidar sensors to detect and avoid obstacles.
LeddarTech Vu8 LiDAR Sensor
The LeddartTech Vu8 is a compact solid-state LiDAR which provides highly accurate multi-target detection over eight independent segments. The Vu8 Lidar sensor weighing only 75 grams can detect obstacles at up to 700 feet (215 meters) range.
The Vu8 uses a fixed laser light source, which significantly increases the sensor’s robustness and cost-efficiency compared with any scanning LiDAR solution.
The Vu8 sensor is very suited for navigation and collision avoidance applications in driver assisted, semi-autonomous and autonomous vehicles such as drones, trucks, heavy equipment for construction and mining, shuttles, buses and other public transportation vehicles.
Applications such as Advanced Traffic Management System (ATMS) requiring longer ranges as well as wide fields of view will also benefit greatly from the new Vu8 sensor offering.
The Vu8 modules are available today in various field of view options at a list price starting at $475, with discount quantities available.
Monocular Vision Sensors For Obstacle Avoidance
Monocular sensors capture images through a single lens camera. It is 3D depth reconstruction from a single still image.
Depth perception is the ability to see things in 3 dimensions and judge distance. As humans we use depth cues when looking at images to determine distances between objects. These depth cues can be binocular or monocular.
Depth cues are also known as Pictorial Depth Cues and there are many of them.
An example of a monocular cue is linear perspective. In a photograph of railway tracks going into the distance, the parallel lines of the track appear to meet together. This gives us the visual perspective of distance.
Another example is when viewing 2 objects which are the same. The object farther away will appear smaller even though the objects are still the same size.
Again, another example of a monocular cue is objects that are farther away appear higher up an image and closer to the horizon line.
Monocular cameras are quite popular and inexpensive. The algorithms used to interpret the image data is what makes monocular vision cameras able to create a 3D images, determine distances between objects and detect obstacles.
In a very simplistic explanation, the algorithm compares the image captured by the monocular vision camera sensor to its pictorial depth cues. That makes it sound very simple. However to achieve obstacle detection using monocular vision cameras has take some outstanding research.
Here is a really terrific article entitled Monocular Visual Mapping for Obstacle Avoidance on UAVs.
Monocular Cameras For Drones
The Parrot AR 2.0 drone has 2 monocular cameras. One front facing and the other downward facing. In fact most drones are equipped with a monocular camera. However, nearly all drones don’t use the monocular cameras for detecting and avoiding obstacles.
However, many researchers are using monocular cameras such as on the Parrot AR 2.0 drone to the detect objects in real time using machine learning algorithms. Here is another article with videos where the Parrot AR drone 2.0 monocular cameras are being used to detect and avoid obstacles using Monocular vision.
Top Drones With Obstacle Avoidance
Lets start with some of the most popular and latest drones with collision avoidance system on the market today.
NOTE: If you buy any drone with obstacle detection and collision avoidance, please read the manual as obstacle avoidance may be switched off in certain modes or will not work if conditions such as lighting or surroundings are not suited to the particular obstacle avoidance sensor.
Here are 2 examples where obstacle avoidance technology may not work. In bad lighting, obstacle avoidance may not work or not work correctly. Certain intelligent modes such as Sports mode on the Mavic, obstacle sensing is switched off.
DJI Skydio 2 Drone With Obstacle Detection And Collision Avoidance
The latest drone on the market, which has obstacle detection and collision avoidance technology is the Skydio 2 drone. It uses a super fisheye lens for 360° view, giving the Skydio 2 true omnidirectional obstacle detection including above and below.
The Skydio 2 was designed specifically for tracking people and objects.
This drone also has exceptional follow you tracking technology and has a fantastic camera allowing you to film stunning 4k video at 60 fps in HDR quality.
The Skydio 2 captures super sharp photos with its 12 MP HDR camera. The Skydio 2 includes many intelligent filming and shooting modes including Dronie, Hover, Angle Track, Orbit and Cable Cam.
The Skydio 2 is manufactured and supported in the US. It’s an true American success story. The founders of Skydio are top tech graduate students from MIT. In 2009, they pioneered autonomous drone technology. After MIT they helped start the Google’s drone delivery program, called Project Wing. Skydio was founded in 2014 and in 2018 they launched the Skydio R1.
Now, in 2019 we have their latest innovative drone, the Skydio 2.
Skydio 2 With Innovative Top Technology
The Skydio 2 is a massive leap in technology from it’s predecessor.
It has a flight time of 23 minutes and is exceptionally light at only 27.3 ounces (775 grams). It can fly at a speed of 36 mph (58 km/h). It has a max ceiling service of 15,000 feet, which is pretty amazing. It also has a transmission range of 2.17 miles (3.5 km).
There are 3 ways to fly the Skydio 2. You can fly using the Beacon, an App on your Smart Phone or by using the Skydio Remote Controller.
Skydio 2 Obstacle Detection Technology
For obstacle detection, the Skydio 2 utilizes 45 megapixels of visual sensing from six x 200 degree color cameras. Skydio 2 visual coverage overlaps and can see everything in every direction with super high resolution and extreme clarity.
To make sense of all the visual data and to react to those visuals, the Skydio 2 has an extremely powerful processing brain. It uses the NVIDIA Jetson TX2, which is the fastest, most power-efficient embedded AI computing device available. With 256 GPU cores, it is capable of 1.3 trillion operations a second. The Skydio 2 really is the flying supercomputer.
Skydio 2 Collision Avoidance Technology
The Skydio Autonomy Engine comprehends what’s happening around the Skydio 2, predicts what will happen next, and makes intelligent decisions multiple times a second. It uses 6 x 4k cameras to build a 3D map of its surroundings, which includes trees, people, buildings and more.
By understanding what you’re doing, what obstacles to avoid and what will happen next, the Skydio 2 will never lose track of its subject.
Skydio Beacon For Tracking
The Skydio beacon gives you unshakably tremendous GPS tracking. It will follow you everywhere, even if it can’t see you. The Skydio Beacon can be used in conjunction with the Skydio app or as a separate, independent device.
When used with the app, Skydio 2 will not be limited by your phone’s WiFi range and will be able to use the Beacon’s strong GPS signal to track its subject.
Skydio 2 Camera
The Skydio 2 has a terrific camera, which is designed around the Sony’s IMX577 sensor and the RedDragon™ QCS605. It is capable of 4k video up to 60 frames per second with HDR.
The Skydio 2, features a dedicated photo mode for capturing 12 MP HDR photos for a variety of shooting options, including single, interval and more.
Here is a video, which shows you all the tremendous features including collision avoidance and follow me mode on the Skydio 2. This drone certainly has the WOW factor.
Kespry 2.0 Collision Avoidance Drone
The Kespry company provide commercial aerial solutions such as inspections and surveying across many sectors including Mining, Telecoms, Construction, Insurance and Roofing.
Their solution includes the Kespry 2.0 drone, the software for mapping and converting images into understandable data for their clients. They also store the data in the cloud for their clients.
Kespry use high resolution cameras such as the Sony UMC-R10C which has a large Exmor APS-C Sensor to capture 20 megapixels of color in detail to accurately calculate volumes, precisely measure distance and angles, and safely identify hazards or damage.
Kespry configures each camera to maximize image quality for specific job types. The geotagged high-resolution images are processed using photogrammetry in the Kespry cloud.
A single orthomosaic image using photogrammetry software is then created to deliver high quality topographic maps, dimensional and volumetric data, and rich business insights.
Kespry 2.0 Collision Avoidance Drone
The Kespry 2.0 drone calculates the flight path and flies autonomously, using LiDAR sensors to avoid obstacles. Kespry don’t go into any real detail about the Lidar sensors which they are using. Overall, if you need a full solution commercial drone, the Kespry 2.0 is worth looking at.
DJI Mavic 2 Pro and Mavic 2 Zoom Obstacle Sensing Drones
The new Mavic 2 Pro and Mavic 2 Zoom just released on 23rd August 2018 can sense objects in 6 directions. Known as Omnidirectional Obstacle Sensing, the Mavic 2 quadcopter can sense objects on to its left, right, up, down, forward and backwards.
Now, you would think that having obstacle sensing on all 6 sides would give you full sensing capabilities. This is not the case. Omnidirectional Obstacle Sensing does not fully cover the full 360 degree circumference arc of the quadcopter.
However, it certainly is the best obstacle sensing drone on the market. There are multirotor drones, which cost 10 times more and haven’t got obstacle sensing on all sides.
How the Mavic 2 Obstacle Sensing Works
The Mavic 2 Pro and Mavic 2 Zoom have Forward, Backward, Downward and Lateral Vision sensors, including Upward and Downward Infrared Sensors. These all provide omnidirectional obstacle detection, providing lighting conditions are adequate.
The main components of the forward, backward and downward vision systems are six camera sensors located on the nose, rear end and underside of the Mavic 2 Pro and Mavic 2 Zoom.
The side Vision system consist of 2 cameras, with one camera on each side of the Mavic 2 quadcopter.
The main components of the Upward and Downward Infrared Sensing Systems are 2 x 3D infrared modules located on the top and underside of the Mavic 2 drone.
The Downward Vision System and Infrared Sensing System allow the Mavic 2 hold its current position and hover in place very precisely. The Vision and Infrared Sensing system allow the Mavic 2 to fly indoors or in other areas where a GPS signal is unavailable.
Notes On Mavic 2 Obstacle Sensing
It is very important to read the documentation which comes with any drone you purchase and especially the sections on safety such as Return To Home features.
If you have a drone with obstacle detection and collision avoidance, it is really important to understand how it works and the situations where it won’t work.
The below notes cover the Mavic 2 but all drones will have similar modes or circumstances where the obstacle detection and collision avoidance won’t work.
ATTI (Attitude) Mode – Obstacle avoidance is not available in ATTI mode as the Mavic 2 Vision System is unavailable.
S-Mode (Sport) – In the Mavic 2 Sports mode, the Vision Systems are disabled and the Mavic 2 only uses GPS for positioning.
RTH Obstacle Avoidance – The Mavic 2 can sense and avoid obstacles during the autonomous return to the Home Point providing that lighting conditions are sufficient and if the GPS signal is strong.
Hyperlapse Intelligent Mode – The Mavic 2 will only brake and hover in place during Hyperlapse.
Parallel & Spotlight Sub Modes – In ActiveTrack 2.0, there are a number of subject modes. In Parallel and Spotlight, the Mavic 2 will brake immediately when an obstacle is detected.
POI Mode – The Mavic 2 will brake and hover in place when an obstacle is detected during Points Of Interest Mode
Lateral Vision Systems – are only available in ActiveTrack 2.0 and Tripod Mode.
Mavic 2 Vision Systems – cannot work properly over surfaces, which do not have clear pattern variations.
Speed – If the Mavic 2 is flying faster than 31 mph (50 km/h), the Vision System does not have enough time to brake and stop.
Too Small of Obstacles – Small obstacles such as leaves or electrical wires are too small to be detected.
Mavic 2 Bottom Auxiliary Light
The Auxiliary light located on the underside of the Mavic 2 improves visibility for the Downward Vision System in bad light conditions.
Notes On Mavic 2 Obstacle Sensing Systems
There are many reasons and environments where the vision and infrared sensing system won’t work or will not work optimally. For example, vision sensors won’t sense in darkness or bad light.
Reflective or transparent surfaces will also cause difficulties with the Vision and Infrared sensors. This is true for all vision systems and not just the Mavic 2.
DJI Mavic 2 APAS System
The Mavic 2 Pro and Mavic 2 Zoom has an Advanced Pilot Assistance System (APAS). The Mavic 2 APAS system, allows users to fly forward and backward avoiding obstacles in front and behind an object.
The APAS will plan an appropriate flight path to avoid these obstacles automatically in flight.
Mavic 2 APAS Notes
- The Mavic 2 APAS feature is automatically disabled using Intelligent Flight modes and resumes after exiting the Intelligent Flight mode.
- It is only available while flying forward and backward. It doesn’t work for sideways flying.
- The Mavic 2 drone will hover in place if it is unable to avoid the obstacle.
- APAS may not function properly over water or snow.
- The Mavic 2 APAS will not function correctly in very dark (< 300 lux) or very bright conditions (> 10,000 lux).
- APAS many not function correctly in No-Fly zones or at Flight limits.
DJI Mavic Air Obstacle Avoidance Drone
The new Mavic Air only released in January 2018 can sense objects in 3 directions. However, its forward and backward vision system is much smarter than other drones. Most of the drones will only hover when it senses and object in front. The Mavic Air is able to recalculate and actually fly around the obstacle. It will only hover if ii cannot remap a route around the obstacle.
The DJI Mavic Air has Forward, Backward and Downward vision system which is constantly scanning for obstacles in front, behind and below. The Forward and Backward vision system allows the Mavic Air to avoid collisions by flying either flying around the obstacle or hovering in front.
FlightAutonomy 2.0 is the name for the DJI integrated system of sensors, algorithms and advanced VIO technology for the Mavic Air. This FlightAutonomy allows the Mavic Air to sense it’s surroundings and take action based on what is senses.
In general terms, VIO or Visual Inertial Odometry technology fuses information from the camera and inertial sensors, specifically IMU, gyroscopes and accelerometers, to accurately estimate device position without relying on Satellite Navigation systems.
Mavic Air Obstacle Avoidance Sensors
The DJI Mavic Air uses advanced VIO technology in its powerful sensor system FlightAutonomy 2.0. This consists of a primary gimbal camera, forward, backward, downward dual vision sensors, downward infrared sensing system, IMU redundancies and a group of computing core processors.
Together, these sensors collect information from the surrounding environment and transmit it to the high performance processor for more precise hovering and better flight performance.
The Downward Vision System helps the Mavic Air to maintain its current location. It can hover in place very precisely. The Mavic Air Downward Vision System also allows the quadcopter to fly indoors or other environments where their is no GPS signal.
The main hardware components of the Downward Vision System are 2 cameras and also a 3D Infrared module.
Mavic Air APAS System
The Mavic Air has an Advanced Pilot Assistance System (APAS) technology which is totally new. The Mavic Air APAS system will allow the quadcopter to attempt to bypass obstacles in front of it, while you are flying using the Remote Controller.
In other words, when you are flying forward, the Mavic Air is constantly scanning it’s surroundings for potential obstacles. If the Mavic Air detects an object or obstacle, it will calculate a safe path around the obstacle without stopping. This is totally new technology which isn’t on DJI other drones, the Mavic Pro or Spark.
If it is unable to calculate or find a safe path, it will stop and hover in place.
DJI Mavic Pro Obstacle Avoidance Drone
Types of obstacle detection sensors used;
- Vision Sensors.
- Ultrasonic Sensors.
This very high tech simple to fly Mavic drone has obstacle detection and collision in 2 directions. It fuses these sensors, its main camera and sophisticated algorithms into its system called FlightAutonomy for obstacle detection and avoidance.
FlightAutonomy is made up of 7 components including 5 cameras (forward and downward dual vision sensors and the main camera), dual-band satellite positioning (GPS and GLONASS), 2 ultrasonic rangefinders, redundant sensors, and a group of 24 powerful, specialized computing cores.
As the Mavic flies, dual forward and downward vision sensors measure the distance between itself and obstacles by taking photos from all four cameras and using the information to create a 3D map that tells it exactly where obstacles are.
The dual forward and downward vision sensors require visible light to function, and in bright light can see as far as 49 feet (15 meters) in front.
The Mavic’s Terrain Follow function uses height information gathered by the on-board ultrasonic system, and its downward facing cameras to keep you flying at the same height above the ground even as the ground moves.
Both the ultrasonic sensors and vision sensors are required to fly indoors or without have GPS signals. The ultrasonic and vision sensors are also used for super smooth hovering and landing.
Here’s a terrific video showing the Mavic Pro obstacle avoidance technology in action. Also notice has smooth the Mavic Pro flies and films. It is also one of the easiest drones to fly.
DJI Phantom 4 Pro Obstacle Avoidance Drone
The Phantom 4 Pro is a truly outstanding drone. It has a terrific 4k camera and flies super stable. In fact, I watched a video recently where one of its propellers was cut off and it still flew very stable. The Phantom 4 Pro also has many auto intelligent flight modes as follows, which make filming real easy;
- Active Track (Profile, Spotlight, Circle).
- Terrain Follow Mode.
- Tripod Mode.
- Gesture Mode.
- S-Mode (Sport).
- P-Mode (Position).
- A-Mode (Attitude).
- Beginner Mode.
- Course Lock.
- Home Lock.
- Obstacle Avoidance.
The Phantom 4 Pro V2 was re-released in January 2020 with some terrific new features. You can read the full DJI Phantom 4 Pro V2 review here, which includes features, specifications and the most frequently asked questions answered.
Phantom 4 Pro Collision Avoidance
The Phantom 4 Pro has 5 directions of obstacle sensing and 4 directions of obstacle avoidance using the following types of sensors;
- Stereo Vision.
- Infrared System.
DJI fuse the above sensors in a system which they call FlightAutonomy.
FlightAutonomy uses high-resolution stereo vision sensors placed at the rear in addition to a pair placed at the front as well as infrared sensing systems placed on the left and right sides. The Phantom 4 Pro uses a 6 camera navigation system.
Three sets of dual vision sensors form a 6 camera navigation system which work constantly to calculate the relative speed and distance between the Phantom 4 and an object.
Using this network of forward, rear and downward vision sensors, the Phantom 4 Pro is able to hover precisely in place without GPS when taking off indoors, on balconies or even when flying through windows with minimal pilot control.
The Phantom 4 Pro is able to fly in complex environments at a flight speed of up to 31 mph (50 km/h) while avoiding obstacles in its path. In Narrow Sensing Mode, the Phantom 4 Pro narrows its sensing scope enabling it to see in more detail and fly through small spaces.
Vision Sensors For Intelligent Flight Modes
Again, DJI use the Phantom 4 vision sensors for more than just collision avoidance. The sensors are used to follow and track objects and people in their autonomous flight modes.
The Phantom 4 is a big choice for many professionals and can be used for a wide variety of uses such as inspections, 3D imaging and professional film making and photography. It is also very competitively priced.
Best Phantom 4 V2 bundle on Amazon.
Phantom 4 Pro V2 direct from DJI.
Here is a terrific video of the Phantom 4 collision avoidance system being tested in its autonomous flight modes. The Phantom 4 has without doubt the best collision avoidance technology when compared to other drones.
Walkera Vitus Collision Avoidance Drone
The new Walkera Vitus fold up drone flies super stable and takes terrific 4k high definition video and 12 megapixel stills. It’s easy to carry and very easy to fly. Walkera always pack loads of technology into their drones.
Walkera Vitus Collision Avoidance System
The Vitus has 3 directions of obstacle avoidance and also sensors for precision hovering. The sensors it uses are as follows;
- Time of Flight Sensor.
- Infrared Sensor.
3 high precision ToF sensors allow the Vitus to detect obstacles 16 feet (5 meters) away in 3 directions (front, left, right).
The Infrared sensor and also an optical flow camera on the bottom of the Vitus takes images at 50 frames per second for positioning and precision hovering. This infrared sensor makes flying indoors possible without satellite signal.
Walkera Voyager 5 Collision Avoidance Drone
The Voyager 5 is the latest professional and commercial drone from Walkera which was only released in early 2018. What I love about Walkera is that they really put as much innovation into their drones as possible.
The Voyager 5 quadcopter integrates many flight safety systems including dual IMU, dual compass, and dual GPS system, in order to make it much more reliable and safer.
The Voyager 5 has a newly designed 3 axis brushless gimbal to enable a more stabilized footage. It uses an advanced shock absorption gimbal technology which greatly reducing vibration and movement during flight, enabling the camera to capture stabilized and fluid footage even after magnifying the focal length.
There are 3 camera options for the Voyager 5 as follows;
- 30x Optical Zoom lens.
- Thermal Infrared camera.
- Low light night vision camera.
Voyager 5 Collision Avoidance Technology
The Voyager 5 has front and downward collision avoidance technology.
The front facing infrared obstacle avoidance module, altimeter and optical flow positioning module, enable the Voyager 5 to better position and sense obstacles while flying, greatly reducing risks caused of crashes.
The Voyager 5 can detect obstacles up to 16 feet (5 meters) in front with a 30° horizontal and ±30° vertical field of view.
The downward vision sensor on the Voyager 5 works from an altitude of less than 10 feet (3 meters). Surfaces should have rich patterns and the lighting should be sufficient.
Yuneec Typhoon H / H Plus Collision Avoidance Drone
This Yuneec Typhoon H and the latest Typhoon H Plus drone uses the Intel RealSense technology to detect and navigate around obstacles. It uses the Intel® RealSense™ R200 camera with an Intel atom powered module to build a 3D model of the world to stop the Typhoon H flying into obstacles. It uses the following sensors;
- Infrared laser camera sensor
- Sonar Sensor
This RealSense technology is capable of remembering its environment, further enhancing the prevention of possible collisions. The Typhoon H collision avoidance system is not reactionary. If it avoids an obstacle once, it will remember the location of the obstacle and will automatically know to avoid it the next time.
The Intel RealSense IR laser camera emits IR light into the scene of where it is going to fly. Based on the displacement of the pattern due to objects in the scene, it can calculate the distance of the objects from the camera. This method to calculate depth in general is known as structured light, and this is the way other 3D cameras, like the original Kinect work.
The intelligent front sonar sensors allow the Typhoon H to stop short of obstacles automatically, ensuring a safer, stress-free flying experience. For example, if the obstacle was too big like a cliff and it can’t avoid it, then the sonar sensors will stop the Typhoon H in front of the cliff.
Obstacle Avoidance In Follow Me Mode
In Follow Me mode, RealSense films in all directions to ensure collisions with objects are avoided. The Intel® RealSense™ R200 camera with Intel® Atom™ powered module builds a 3D model of the world, allowing you to focus on the subject without worrying about flying into obstacles.
DJI Matrice 200 Collision Avoidance Drone
The DJI Matrice 200 is the latest commercial drone from DJI and has many uses including inspections of power line, bridge, cellphone towers etc. It is very adaptable and can carry the Zenmuse X4S, X5S, Z30 and XT cameras.
It can also carry a camera on top of the quadcopter and 2 cameras under the Matrice 200. So you have a Zenmuse Z30 zoom camera and a thermal vision camera mounted under the drone.
The Matrice 200 has many dual systems for fail safe redundancy such as dual battery, satellite navigation, IMU and IP43 protection. It also has many intelligent flight modes such as Points of Interest and ActiveTrack.
DJI M200 Obstacle Detection And Collision Avoidance
For obstacle detection and collision avoidance, the DJI Matrice M200 combines various sensors as follows;
- Time of Flight laser Sensor.
- Stereo Vision Sensor.
- Ultrasonic Sensor.
An upward facing Time-of-Flight laser sensor camera recognizes objects above. The Matrice 200 uses Stereo Vision sensors to detect objects in front. It also uses both Stereo Vision and Ultrasonic sensor below.
The Vision System consist of 3 stereo vision sensors and 2 ultrasonic sensors on the front and bottom.
There are 2 Time of Flight infrared laser sensors on top of the Matrice 200.
This whole Vision System constantly scans for obstacles allowing the Matrice 200 to go over, around or just hover in front of the obstacle.
DJI Inspire 2 Collision Avoidance Drone
The DJI Inspire 2 is a dream come true for professional filmmakers and cinematographers as it supports many features that demanding aerial videographers require, including dual-operator control and pro-grade video compression.
The Inspire 2 has added features to increase its reliability with dual redundancy of key modules such as the IMU and barometer. The intelligent flight control system monitors the redundancy system, giving it accurate flight data.
To make filming a lot easier, the Inspire 2 has the following intelligent flight modes;
- Spotlight Pro.
- Profile mode.
- Tripod mode.
All these intelligent flight modes make the Inspire 2 very easy to fly and allowing and give the pilot the ability to concentrate to create complex, dramatic shots.
The quality of film produced by the Inspire 2 would make a Hollywood producer proud. Integrated into the Inspire 2 is the brand new CineCore 2.1 image processing system, capable of recording 5.2 and 6k videos in CinemaDNG, 5.2k videos in Apple ProRes and more. CineCore 2.1 is built into the aircraft nose and works with any camera connected through the dedicated gimbal port.
The Inspire 2 is the only drone you will need for producing Hollywood quality film. It comes with ground gimbals so you can film 100% of the movie or documentaries with the Inspire 2. In fact, a film called The Circle was filmed 100% with the Inspire 2. The quality of these film is stunning.
DJI Inspire 2 Obstacle Avoidance System
The Inspire 2 uses the following sensors in its Vision and Infrared Sensing system to sense and avoid obstacles;
- Stereo Vision Sensors.
- Ultrasonic Sensor.
- Infrared Sensor.
The main components of the Vision System is on the front and bottom of the Inspire 2 which includes 2 x stereo vision sensors and 2 x ultrasonic sensors.
The Infrared sensing system consists of 2 x Infrared modules on the top of the Inspire 2.
The forward and downward vision systems enable the Inspire 2 to detect obstacles up to 98 feet (30 meters) ahead, allowing for protected flight at up to 34 mph (54 km/h) at a controllable attitude angle of 25°.
The upward facing infrared sensors scan obstacles 16 feet (5 meters) above, adding protection when flying in enclosed spaces. Obstacle sensing systems are active during normal flight, Return To Home and all Intelligent Flight Modes.
The DJI Inspire 2 is the drone we all dream of. You can read more on the fantastic features of the Inspire 2 in this terrific DJI Inspire 2 review.
Below we have a nice graphic which shows where the DJI Vision System sensors are located on the Inspire 2.
Autel Evo Obstacle Detection
The Autel Evo aerial photography drone is a very easy to fly quadcopter, even indoors or at low altitudes. The EVO has a tremendous flight time of 30 minutes and a 4.3 mile (7 km) video range.
With Dynamic Track, Obstacle Avoidance and 3D Mapping technology, EVO takes safety and stability seriously.
EVO includes a remote controller which houses a 3.3-inch OLED screen providing you with critical flight information or a live 720p HD video feed letting you see the camera view without the need for a mobile device.
Autel Evo Obstacle Detection And Collision Avoidance
Utilizing two cameras on the front giving it binocular vision EVO creates a 3D environment and reacts to obstacles in the way. Intelligent algorithms are constantly running during autonomous flight, making long-range decisions for path planning around obstacles.
Two ultrasonic sensors paired with two more computer vision cameras on the bottom of EVO, help protect the aircraft from landing on unlevel surfaces. Using the precision landing feature the cameras on the bottom of EVO will capture reference images and use them during the return to home providing pinpoint precision during the landing sequence.
On the rear of the aircraft, EVO is equipped with a near IR sensor protecting you when flying backward autonomously.
Autel Evo Aerial Drone Camera Specifications
The Autel Evo is equipped with a powerful aerial camera on a 3-axis stabilize gimbal, which records video at 4k resolution up to 60 frames per second and a recording speed up to 100 mbps in H.264 or H.265 codec. Using real-glass optics EVO captures stunning aerial photos at 12 megapixels with a wide dynamic range for more details and color.
- Resolutions: 4k/12 MP camera.
- FOV: 94°.
- Video: 60 FPS video.
- Sensor: Sony CMOS 1/2.3″ sensor.
- Aperture: F2.8.
- Image Processor: Ambarella H2.
- Supported SD Card Types: Micro-SD Card up to 128 GB Class 10.
- File Formats: Photo: JPG, RAW, JPG+RAW.
Autel Evo Aerial Camera Modes:
- Single shot.
- AEB – 3/5.
- Burst shooting – 3/5/7/14.
- Time lapse – 2/5/7/10/20/30/60.
Autel Evo Video Resolution
- 4k 3840 x 2160.
- 4k+ 4096 x 2160.
- 2.7k 2720 x 1530.
- 1080P 1920 x 1080.
- 720P 1280 x 720.
Video Frame Rate: 240 FPS, 60 FPS, 48 FPS, 30 FPS and 24 FPS.
Not to finish, here is the launch of the Autel Evo quadcopter. It sure is a fantastic obstacle detection quadcopter.