Unmanned Systems Technology 002 | Scion SA-400 | Commercial UAV Show report | Vision sensors | Danielson Trident I Security and safety systems | MIRA MACE | Additive manufacturing | Marine UUVs

67 MIRA MACE | Dossier mechanical system into 32 beams that scan the area in front of the vehicle, capturing any reflections in a sensor. Lidar sensors don’t lend themselves to military applications as they emit so much laser light that they show up clearly in the dark, making them an obvious target. They have been used though for optical mapping of an area, with a remote operator driving the vehicle, and the map is then used to define the waypoints for autonomous operation. They are also used for range finding as they provide a longer range than optical cameras, detecting objects at 50-85 m. All the data is fed into a ruggedised laptop running a customised version of Linux that is optimised for the control of the system. The critical part of the obstruction detection algorithm kicks in at 30 m to interface with the image detection to the VIM, triggering the throttle and braking algorithms simultaneously. The detection algorithm then assesses whether it is an obstacle or terrain, using the image data rather than data from the engine and transmission. The core of the control algorithm is a fusion of data from the camera systems and GPS so that a series of GPS tracks can be pre-loaded over a digital map, but how it controls its speed is based on the optical systems. As a result, MACE 3 will know how to drive that route a lot quicker than in remotely operated mode as it has more precise data from the cameras and so can stop faster than a remote driver could. In the self-driving mode it is driving according to the parameters of a route, so it won’t deviate from its route but will vary its behaviour. The same approach is used for the cooperative driverless car that MIRA has developed. This takes its position and navigation cues from messages sent from other cars and from roadside sensors. For MIRA the driverless car is an easier thing. The whole point of the military application is that a situation can be guaranteed to go wrong, and the environment can and will change dramatically. This means the key is having redundancy in the design of the system so you can throw more variables at it. According to the development team, this in turn means MACE 3 is able to make all the decisions that the cooperative car can make but in a much more random environment. Interestingly, MIRA is conducting research into a road-edge detection system for off-roading, the idea being to take a snapshot 15 m ahead that interprets a road edge. That will mean being able to navigate along tracks in a much more autonomous fashion rather than a pre-recorded track, and this is all down to the signal processing algorithms used. Security Because MACE 3 is a military design, system security has been built in from the start, and accounts for most of the complexity of the design. This is especially the case with the comms system, which has to be agile in frequency, have long range and be secure from hacking and jamming. But there are so many sensors and systems that are part of an autonomous vehicle that the number of vectors by which a cyber attack could be directed on an autonomous or connected vehicle is myriad, says MIRA. However, because the MACE platforms are designed to interpret things that are unpredictable, they will be harder to attack. The pinch point on most of these platforms comes down to comms and the data link – this is invariably the weak point and is where it can get expensive, as the system is only as good as the comms, which is where MIRA spends a lot of time to make it work. Conclusion The experience of the MACE and Panama autonomous platforms over the past decade has invaluable lessons for the development of the driverless car. The implementation of the control algorithms, fusion of camera data with other sensors in a simple and efficient manner, and the integration of the comms systems are all elements that have been tested in the most hostile environments possible. The technology developed for these environments will translate to more efficient and effective control systems for autonomous, connected vehicles on the road. Unmanned Systems Technology | Spring 2015 The Motor Industry Research Institute (MIRA) was founded in 1946 as the UK government’s automotive research centre. In the past 60 or so years it has built crash labs and system test capabilities, and developed new technologies such as one of the first production electric cars, the Enfield, in 1972. In 1982 it opened its Engine Test Laboratory, initially with six engine test cells. The facility has since been expanded with the addition of a further eight test cells. MIRA has a Technology Park, a 1.75 million sq ft enterprise zone, in the Midlands region of England, and the latest company to take a bespoke facility there is Aston Martin, whose 15,800 sq ft building will accommodate a key part of its prototype and vehicle development. Aston Martin joins other companies who have already located at the park (which was awarded Enterprise Zone status in 2011), including Haldex, Bosch, Norgren and Lockheed Martin. The aim is to house businesses from across the automotive, aerospace, rail, defence and associated digital technology sectors. About MIRA