Low-Power Real-Time Planning for Robots in Uncertain Environments

NASA's Mars Sample Return (MSR) mission requires a team of autonomous robots---a rover, lander, and orbiter---to work in concert to effectively return samples from the Martian surface to Earth. These robots must be able to work together, autonomously, in real time: transmission limitations effectively preclude coordination with mission control. Even worse, each of these robots must plan with severely restricted computational abilities. The central objective of this proposal is to improve the capabilities of complex, autonomous robots that must operate in real-time conditions in partially unknown environments, be tolerant of failure, and do so with heavily constrained computational resources. First, this work will focus on accelerating common planning subroutines to increase the efficiency of conventional planning algorithms. This will be done by taking advantage of CPU single-instruction-multiple-data (SIMD) parallelism, among others. The already visible target is to obtain planning in microseconds, which will be two orders of magnitude faster than today's motion planning. The second aim of the project will ensure that such a capability is retained in dynamically changing environments by effectively connecting such planners' input to raw sensor data, such as point clouds. Then, as a third aim, risk-aware planning will be investigated. The goal here is to take advantage of proposed accelerations to produce new trajectories at near-control-loop speeds. The new paradigm will challenge conventions on how planning is used in uncertain environments, when paths are repaired and when they need to be re- computed, and how learning-based algorithms for planning can be exploited. The usefulness of the proposed framework is not limited to the planning of one robot but many, enabling multi-robot teams to coordinate cooperatively, eventually, with little help from a centralized operator. This last aim of the project will bring together all the developed capabilities to deliver robotic collaboration for complex tasks. This work will advance NASA's "Explore: Autonomous Systems and Robotics strategic" thrust---specifically the technology objectives for "self-adaptive and fail-active autonomy for high-tempo missions" as well as "efficient on-board autonomy for continuous surface operations"---by improving the capabilities of autonomous robots to plan on-line without human intervention, tolerate failure, and operate continuously for long periods of time.

This is early/mid-stage (TRL 2) — the most realistic path in is NASA SBIR/STTR, which funds small businesses and research institutions to develop technology aligned with NASA's needs (equity-free, phased funding). Check whether a current SBIR/STTR solicitation topic overlaps with this project's technology area, or contact the project directly (above) to ask.

None of these are guaranteed paths for this specific project — TechPort itself doesn't have an "apply" button. Reaching out to the contact(s) above with a specific question is usually the fastest way to find out what's actually open.