Dexterous manipulation strategies implemented on the Delta Arrays using various implementations.
- Model Free Reinforcement Learning with Dynamic Motion Primitives
- Open Loop Motion and Contact Planning Algorithm - HiDex
- Closed Loop Visual Feedback with Causal Inference and Probabilistic Graphical Models for Planar Manipulation
- Dexterous Distributed Manipulation using SAC and Multi-Agent RL
- (new) A Diffusion-based Policy to generate object trajectories.
This is a research project to aid development of modular sample-efficient dexterous manipulation algorithms that are benchmarked with an array of 64 Delta Robots in an 8x8 grid.