Terrain-Aware Control and Model-Driven Learning for Quadrupeds in Low Gravity

In the coming years, we look towards the stars and aim toward becoming a multi-planetary species. The return to the moon hasalways been forthcoming since our last encounter in 1969, and with the recent discovery of water on the sunlit side of the moonby NASA’s SOFIA, this mission becomes ever critical. We also look farther into the solar system at similar bodies such as Marsand Jupiter's moons. However, the journey to these bodies and setting up a habitable colony will not be easy. The terrains in theseinterplanetary entities will be vastly different than Earth’s terrain and will feature craters, ice, and even mountainous grounds.Moreover, these habitable entities typically have lower gravity than on Earth. The two features will make locomotion by humansand their assistive robots particularly difficult. Firstly, the low gravity makes for less traction and the uneven terrain would becumbersome to wheeled robots. As such, we envision that legged robots, more so quadrupeds, will be at the astronaut’s side to help setup habitable colonies on other worlds.Robots that will aid the exploration of interplanetary environments will require fast and fluid motions and must be able to transverseterrains that vary from sand to ice or even mountainous grounds. To achieve this feat, robotics control solutions, e.g., optimal control(OC) methods, will need to be a) faster than real time and b) aware and responsive to the low gravity problem.In this research proposal, we aim to answer these challenges via a set of four objectives. 1) Firstly, we will develop control strategiesfor quadrupeds to navigate complex and varied terrains. Here, we will use a tool named Differential Dynamic Programming (DDP)that can produce OC solutions in real time. This tool is extremely powerful and features a correction term that enables maintainingstability without re-optimizing for quadrupeds in different terrains. Here, we note that DDP relies on an initial trajectory in itsoptimization. The choice of this initial trajectory is critical to the computation speed of DDP.This observation brings about our second objective. 2) This research will pursue the development of a learning library of gaits,footholds, terrains, and their correspondent OC. Once developed, when the quadruped encounters a terrain in the other world, itwill query the library for a trajectory. It will use the queried trajectory as the initial guess. This will allow for faster than real-timecomputation of optimal gaits in different terrains. Moreover, this learning library will be made to be dynamic so that it grows overtime (within limits) to add new terrain information. These two objectives deal with the first problem of computing control laws fasterthan real time, but we also need to deal with the problem of low gravity.To deal with this problem of low gravity, we introduce the third objective. 3) This research objective will consider the addition of anextra appendage (flywheel or control moment gyroscope) to the quadruped. This extra appendage will generate the requisite bodytorque to counteract the reduced traction and thus will help to maintain stability in low-gravity environments. The control problemis now seen through the lens of handling the conservation of angular momentum. This advance will lead to a quadruped that is moreresponsive in low-gravity environments and even in the presence of external disturbances.Finally, the last objective will (4) validate and unify all the previous objectives in simulations and hardware experiments. We willfocus on quadruped robots like the University of Notre Dame’s MIT mini-cheetah and potentially other mobile platforms such as theNASA Jet Propulsion Laboratory's (JPL) LLAMA and LEMUR multilegged robots. The visiting technologist role will serve as anopportunity to inject relevant data into the project, carry out experiments with new systems, and work on the outlined objectives.

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.