Develop concepts or technologies for new sensors, sensing techniques, perception, and intelligence algorithms to generate actionable information for an autonomous vehicle to successfully move through a rugged off-road environment.
Autonomous off-road mobility is becoming a critical requirement for various military operations, including maneuvering in rugged environments and disaster relief. Vehicles in these missions will require sensors and software to navigate unexpected obstacles and have the hardware necessary to interact with these potentially difficult elements. Some of the challenges autonomous vehicles may face include objects that do not inhibit forward movements but occlude or confuse sensors, such as tall grass, brush, or debris. Similarly, other obstacles, such as fences or glass, may not be easily detectable given existing technology.
These systems should also be built around the fundamentals guiding trusted or assured autonomy, which encourage developing technologies that can operate with minimal human intervention. These systems must be designed to surmount potentially more challenging scenarios than simple navigation challenges and may require higher-level cognition.
Obstacle Detection and Assessment
Vehicle sensors should predict if and how objects (e.g., tree trunks, branches, bushes, tall grass, rocks) will move if the vehicle exerts a force on them. These systems should also account for difficult-to-detect objects of importance (e.g., wires, fences, glass).
- Are existing sensors up to the task of doing obstacle detection, or are there potential enhancements that could improve their accuracy?
- How could algorithms be designed in a way to improve sensor pickup?
Autonomous ground vehicles may be required to operate in environments without access to GPS or other sensing systems. Concurrently, they may need to limit their sensor signatures for security reasons.
- What sorts of technologies could assist ground vehicles in conducting passive odometry without significant sensor emissions?
- Could ground vehicles communicate with one another – or a human – using alternative methods?
Given the many risks associated with operating autonomous systems, designing these vehicles to ensure safety and reliability is essential.
- What sorts of subroutines need to be designed to make autonomous vehicles function independently and potentially control other vehicles in the field?
- What sorts of feedback mechanisms can help appropriately calibrate trust in the autonomous system across various tasks and conditions?
- Is there a way to prevent autonomous systems from going rogue because of adversary interference?
Awards:- Innovators work to develop solutions for autonomous vehicles