ManiWAV
Robot control for learning outdoor audio and visual data
- Provide rich interaction and object attribute information through audio signals
- Synchronize audio and visual feedback collection using an 'ear in hand' data collection device
- Learn robot control strategies directly from human demonstrations
- Showcase system capabilities in four diverse control tasks
- Realize generalization of unseen environments through diverse field human demonstration learning
Product Details
ManiWAV is a research project aimed at learning robot manipulation skills through outdoor audio and visual data. It collects synchronized audio and visual feedback from human demonstrations and learns robot control strategies directly from the demonstrations through corresponding strategy interfaces. This model demonstrates the ability of its system through four contact rich manipulation tasks, which require the robot to passively perceive contact events and patterns, or actively perceive the materials and states on the surface of objects. In addition, the system can generalize to unseen wilderness environments by learning from diverse human demonstrations in the wild.
