Additional Funding Sources
This research was supported by NASA via Idaho Space Grant Consortium.
Abstract
Marsha (Multi-Agent Reinforcement Self-supervised Heuristic Algorithm) is an artificial intelligent robotic system that learns to play catch in space using a combination of multi-agent communication, meta-learning, computer vision, and reinforcement learning techniques. Marsha's ability to adapt to unseen environments will lead to advances in manufacturing, repair and autonomous control on Earth and in Space. The robotic arm features tendon driven mechanical hands for cooperative dexterity manipulation, a neural network driven control system, and a space grade soft robotic gripper which are currently being researched. The demonstration of these technologies is scheduled to launch on a sounding rocket from the Wallops Flight Facility in August 2022 as part of the Colorado Space Grant Consortium's RockSat-X Program.
Multi-Agent Artificial Intelligence for Adaptive Robotics
Marsha (Multi-Agent Reinforcement Self-supervised Heuristic Algorithm) is an artificial intelligent robotic system that learns to play catch in space using a combination of multi-agent communication, meta-learning, computer vision, and reinforcement learning techniques. Marsha's ability to adapt to unseen environments will lead to advances in manufacturing, repair and autonomous control on Earth and in Space. The robotic arm features tendon driven mechanical hands for cooperative dexterity manipulation, a neural network driven control system, and a space grade soft robotic gripper which are currently being researched. The demonstration of these technologies is scheduled to launch on a sounding rocket from the Wallops Flight Facility in August 2022 as part of the Colorado Space Grant Consortium's RockSat-X Program.