Goal-Conditioned End-to-End Visuomotor Control for Versatile Skill Primitives

Groth, Oliver; Hung, Chia-Man; Vedaldi, Andrea; Posner, Ingmar

doi:10.48550/arxiv.2003.08854

Cited by 1 publication

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Experimental Setup and Data Collection. We follow Gorth et al [31] and use the MuJoCo physics engine [32] along with an adapted Gym environment [33] provided by [4] featuring the Fetch Mobile Manipulator [34] with a 7-DoF arm and a 2-finger gripper. Three tasks (Figure 3) are designed as they are fundamental in manipulation and commonly used as building blocks for more complex tasks.…”

Section: Methodsmentioning

confidence: 99%

Introspective Visuomotor Control: Exploiting Uncertainty in Deep Visuomotor Control for Failure Recovery

Hung¹,

Li²,

Wu³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

End-to-end visuomotor control is emerging as a compelling solution for robot manipulation tasks. However, imitation learning-based visuomotor control approaches tend to suffer from a common limitation, lacking the ability to recover from an out-of-distribution state caused by compounding errors. In this paper, instead of using tactile feedback or explicitly detecting the failure through vision, we investigate using the uncertainty of a policy neural network. We propose a novel uncertainty-based approach to detect and recover from failure cases. Our hypothesis is that policy uncertainties can implicitly indicate the potential failures in the visuomotor control task and that robot states with minimum uncertainty are more likely to lead to task success. To recover from high uncertainty cases, the robot monitors its uncertainty along a trajectory and explores possible actions in the state-action space to bring itself to a more certain state. Our experiments verify this hypothesis and show a significant improvement on task success rate: 12% in pushing, 15% in pick-and-reach and 22% in pick-and-place.

show abstract

Section: Methodsmentioning

confidence: 99%