Artificial Intelligence for Prosthetics - challenge solutions

02/07/2019

∙

In the NeurIPS 2018 Artificial Intelligence for Prosthetics challenge, participants were tasked with building a controller for a musculoskeletal model with a goal of matching a given time-varying velocity vector. Top participants were invited to describe their algorithms. In this work, we describe the challenge and present thirteen solutions that used deep reinforcement learning approaches. Many solutions use similar relaxations and heuristics, such as reward shaping, frame skipping, discretization of the action space, symmetry, and policy blending. However, each team implemented different modifications of the known algorithms by, for example, dividing the task into subtasks, learning low-level control, or by incorporating expert knowledge and using imitation learning.

READ FULL TEXT

Artificial Intelligence for Prosthetics - challenge solutions

Learning to Run challenge solutions: Adapting reinforcement learning methods for neuromusculoskeletal environments

Fuzzy Controller of Reward of Reinforcement Learning For Handwritten Digit Recognition

Learning Algorithms for Regenerative Stopping Problems with Applications to Shipping Consolidation in Logistics

ProLoNets: Neural-encoding Human Experts' Domain Knowledge to Warm Start Reinforcement Learning

Learning the aerodynamic design of supercritical airfoils through deep reinforcement learning

Reward is not enough: can we liberate AI from the reinforcement learning paradigm?

Towards sample-efficient episodic control with DAC-ML

Artificial Intelligence for Prosthetics - challenge solutions

Related Research

Learning to Run challenge solutions: Adapting reinforcement learning methods for neuromusculoskeletal environments

Fuzzy Controller of Reward of Reinforcement Learning For Handwritten Digit Recognition

Learning Algorithms for Regenerative Stopping Problems with Applications to Shipping Consolidation in Logistics

ProLoNets: Neural-encoding Human Experts' Domain Knowledge to Warm Start Reinforcement Learning

Learning the aerodynamic design of supercritical airfoils through deep reinforcement learning

Reward is not enough: can we liberate AI from the reinforcement learning paradigm?

Towards sample-efficient episodic control with DAC-ML