We provide fully-autonomous navigation and route planning. You don’t have to spend your time specifying dozens of GPS waypoints for your drone to blindly follow. Simply specify the areas you need data from. Our drones will figure out an efficient route to cover the areas and collect the data you want.
You don’t need an expert pilot. We provide active obstacle detection and avoidance capabilities. Our drones not only avoid the obstacles they encounter, but continuously plan how to safely navigate around them and get to where they need to go.
Our software provides a number of safety features to ensure that the drones operate safely and predictably.
A geo-fenced mission area is required for every flight. The drone is not allowed to leave the mission area for any reason. The drone will respect the mission area boundaries when planning its route to cover the specified areas of interest and when planning its path to its current target and around any obstacles. If the drone finds itself outside the mission area during the flight for any reason (strong winds blowing it off-course, malfunction, etc.), an automatic landing procedure is triggered. The drone is disabled and not allowed to take off when outside the mission area.
Our system continuously monitors itself to ensure all sensors and devices are operating normally. If a system fault is detected, the drone will initiate an immediate landing or a return to the nearest base, depending on the severity of the fault.
The remaining battery level and estimated remaining flight time are continuously monitored and updated. The drone will suspend its mission and return to the nearest base when the battery level is low. If the battery level drops below a critical threshold and does not allow reaching the nearest base, the drone will initiate an immediate landing to avoid running out of power in mid-air.
Our software allows for remote operator intervention. The operator can initiate a return to the nearest base, an immediate landing, or even an emergency stop (stopping the motors) if they feel the drone is behaving unsafely.
Obstacle Detection & Avoidance
Our software uses a stereo camera as an inexpensive and reliable sensor for extracting 3D information about the world. The stereo camera provides a continuous stream of synchronized pairs of images from two different vantage points. Using the differences between each pair of images, we extract a dense 3D point cloud of the drone’s surroundings. We process the point clouds in real-time and use them to build and update a 3D map of the environment. The map keeps track of free, occupied, and unobserved space. The drone is only allowed to fly through free space and needs to maintain a minimum safety distance from any occupied space.
Instead of simply avoiding the occupied space, the drone continuously plans how to safely navigate the environment to reach its goal. The path planner also reasons about how to orient the stereo camera in order to observe any unobserved space around the drone before flying through it.
Our software can operate with a wide variety of aerial vehicles providing unparalleled flexibility. We can incorporate a myriad of sensors and payloads specific to our customers’ needs. High resolution cameras, thermal imaging cameras, pollution sensors, and many more–if it can be carried by a drone, we can use it to collect data. Our goal is to provide the right tool for the job at hand. Let us help you choose the right drone at the right price for your needs.
AI & Machine Learning
Our software uses state-of-the-art artificial intelligence and machine learning to recognize, locate, and collect the right data. We use neural networks to identify and categorize a variety of objects–people, vehicles, buildings, solar panels, and many others. With the right data, we can teach our drones to perform a wide variety of tasks–locating and following objects of interest, performing inspections, mapping and surveying. The more they fly, the more they learn.
Designed for Autonomous Operation
Our software was designed for fully autonomous operation from the very start. The autonomous navigation system is the cornerstone of our technology. Our reliable, safe, and robust autonomous flying system allows us to easily build a customized solution for almost any task. We think of it as our “expert pilot”. It just needs to be paired with the right “navigator” to tell it where to fly to and what to look for. Our route planning system is an intelligent “navigator” that can calculate efficient routes for covering areas of interest and visiting points of interest. The “navigator” uses AI and machine learning to detect and locate objects of interest. It knows what to look for and what to do when it finds it. It is flexible and easily programmable, which allows us to develop customized solutions very quickly.
Overcoming Roadblocks & Limitations
Our goal is to provide you with the fastest and most reliable tool for your data collection needs. We aim to eliminate wasted time and get you the data as quickly and efficiently as possible. We can’t control the weather, but when it is right for flying, our autonomous “expert pilot” and “navigator” will be there and ready to go.
We are aware that the current battery technology imposes limitations to the range and flight time of any drone. Our system was designed to suspend and resume missions to allow for changing of batteries. The progress along the mission is stored in non-volatile memory allowing it to be resumed even after a cold shutdown.
Accelerating Development & Testing
To accelerate development and testing of new features, we use a realistic physics simulator. The simulator allows us to do fast initial testing and catch issues early, without worrying about damaging expensive hardware. It also allows us to do much more extensive testing, verification, and validation of our software without weather or environmental delays. We can perform hundreds of simulated flights per day in different environments and conditions to ensure our drones perform safely and reliably. The simulator also allows us to simulate rare events, such as hardware failures, which are difficult to encounter during real-world testing.