Learning to Generate Cost-to-Go Functions for Efficient Motion Planning

Huh, Jinwook; Xing, Galen; Wang, Ziyun; Isler, Volkan; Lee, Daniel D.

doi:10.48550/arxiv.2010.14597

Cited by 1 publication

(2 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…By generating cost-to-go functions over the entire C-space by HOF, we overcome sample complexity issues and directly generate efficient plans for various environmental conditions. We emphasize that the present architecture is a significant improvement from earlier work [4] which required the collision map in the configuration space to first be constructed.…”

Section: Related Workmentioning

confidence: 99%

“…In previous work [4], we investigated the problem of generating the cost-to-go function over the configuration space where we used an explicitly constructed configuration map (C-map) as an intermediate step. In other words, for a d DoF robot, the approach in [4] created a d-dimensional binary map representing collisions which was then transformed into a cost-to-go map for a fixed destination. This approach worked well for low DoF robots where the C-map can be represented explicitly on a grid.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Cost-to-Go Function Generating Networks for High Dimensional Motion Planning

Huh¹,

Isler²,

Lee³

2020

Preprint

Self Cite

View full text Add to dashboard Cite

This paper presents c2g-HOF networks which learn to generate cost-to-go functions for manipulator motion planning. The c2g-HOF architecture consists of a cost-togo function over the configuration space represented as a neural network (c2g-network) as well as a Higher Order Function (HOF) network which outputs the weights of the c2g-network for a given input workspace. Both networks are trained end-to-end in a supervised fashion using costs computed from traditional motion planners. Once trained, c2g-HOF can generate a smooth and continuous cost-to-go function directly from workspace sensor inputs (represented as a point cloud in 3D or an image in 2D). At inference time, the weights of the c2g-network are computed very efficiently and near-optimal trajectories are generated by simply following the gradient of the cost-to-go function.We compare c2g-HOF with traditional planning algorithms for various robots and planning scenarios. The experimental results indicate that planning with c2g-HOF is significantly faster than other motion planning algorithms, resulting in orders of magnitude improvement when including collision checking. Furthermore, despite being trained from sparsely sampled trajectories in configuration space, c2g-HOF generalizes to generate smoother, and often lower cost, trajectories. We demonstrate cost-to-go based planning on a 7 DoF manipulator arm where motion planning in a complex workspace requires only 0.13 seconds for the entire trajectory.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%