Oxbotica launches Selenium mobile autonomy software
23 July 2016
Oxbotica, a spin-out from Oxford University’s Mobile Robotics Group, launched its new Selenium mobile autonomy software solution with a purpose-built concept vehicle named Geni.
Selenium can work in pedestrianized environments as well as roads and motorways, and is not reliant on GPS to operate—i.e., it can easily transition between indoor and outdoor settings, over ground or underground. The system has been developed to be “vehicle agnostic”—it can be applied to cars, self-driving pods (e.g. for campuses and airports), and warehouse truck fleets.
The software developed by Oxbotica’s team is able to provide any vehicle to which it is applied with an awareness of where it is, what surrounds it and, with that knowledge in hand, how it should move to complete a task.
Selenium’s system uses patented algorithms that gives vehicles intelligence autonomously to perform a range of mobility tasks, including motion control, braking, calibration, navigation, static and dynamic obstacle detection.
The patented integrated laser- and vision-based, GPS-free approach produces 3D models of large-scale environments quickly, cheaply and robustly. These highly accurate maps enable vehicle localization, infrastructure-free navigation, scene perception and intelligent self-driving vehicles.
The same 3D maps can be the foundation blocks for multiple applications ranging from Smart City planning, survey, infrastructure management, asset detection/management and low-cost localization for devices ranging from smartphones to cars, buses and trains.
| Oxbotica has developed technology for automatically detecting and removing 3D objects from point clouds. This video demonstrates how this might be used for the detection of cars in an outdoor scene with the ultimate aim of removing non-permanent features. |
Selenium will be deployed at a series of autonomy trials in which Oxbotica is the sole supplier of autonomy software, including the £8-million (US$10.5-million) GATEway project in Greenwich and the LUTZ Pathfinder self-driving pod project in Milton Keynes.
In the GATEway project, Oxbotica is providing the Selenium autonomous control system to power 8 shuttle vehicles developed by Westfield Cars and Heathrow Airport, which will be used by members of the public in a 6-month demonstration starting mid-2016. The company is also providing the Caesium shuttle management system—a cloud-based service that will schedule and co-ordinate the small fleet of autonomous vehicles, enabling smartphone booking, route optimization and data exchange between the vehicles without human intervention.
The company is also working with manufacturers in a broad spectrum of mobile autonomy domains, driverless cars.
Oxbotica has more than 70 modular pieces of mobile-autonomy-related IP which range from patents, to software, to “know how” which can be integrated into customers’ products and applications to enable end-to-end autonomy solutions.
Another important development for future (independent) automated drive vehicles.
An ADV not using GPS and/or outside data would probably be safer and less affected by interferences and weather conditions.
Combined with (some) existing sensors + data gathering + treating systems, a safer ADV may become a reality.
Posted by: HarveyD | 23 July 2016 at 11:22 AM
It will be interesting to see if 3rd party AV software is developed (like Android for phones). It would be really handy for car manufacturers did not have to write their own AV s/w as it is a huge, specialised task.
It should also be portable, much in that way that a human can get into any rental car and drive it off. Once you can observe the road, all you really have to do is go left, go right, speed up or slow down and emergency stop.
You would need a set of sensors, probably stereo cameras, lidar and radar and reversing sensors (and GPS) - nothing too hard here (assuming the lidar cost can be kept low enough).
You would want standard interfaces to the sensors and a standard interface to the engine, brakes and steering, and off you go.
I see no reason for each car manufacturer to develop their own. If you have 2 or 3 decent systems, you should have a reasonably efficient market.
Posted by: mahonj | 25 July 2016 at 02:56 AM