Event
Zhejiang University and TeleAI propose GR-BC robot training method
1 reports · 1 sources
The team led by Shen Chunhua at Zhejiang University, together with the China Telecom Artificial Intelligence Research Institute (TeleAI) and Bai Chenjia's team at Gamma Intelligence, proposed the GR-BC method: multiple attempts at the same task are constructed into a graph, reverse Dijkstra search is used to estimate the remaining number of steps to success from each state, and action chunks are then weighted accordingly to train a vision-language-action model. The main experiments use X-VLA as the backbone; each method starts from the same behavior cloning checkpoint, and for each task the evaluation uses a fixed set of 30 successful and 30 failed trajectories; the team reports reaching 91% on the four X-VLA tasks. The work has been accepted to CORL 2026, and the paper and code have been made public.
Report timeline
- Zhejiang University and TeleAI propose GR-BC, using graph search to select effective actions from robot trial and error
The team led by Chunhua Shen at Zhejiang University, together with the China Telecom Artificial Intelligence Research Institute (TeleAI) and Chenjia Bai's team at Gamma Intelligence, proposed GR-BC, which builds multiple attempts at the same task into a graph, uses reverse Dijkstra search to estimate the remaining steps from each state to success, and then weights action chunks accordingly to train a vision-language-action model.
机器之心 · 微信公众号 ·
Related topics
Interest trend
- Current
- 3
- Peak
- 4Nov 4 01:00
- Past 24 hours
- –
The trend compares participants observed throughout the same period, so its range may be smaller than the current interest count. Move or tap to inspect each hour; use the arrow keys on a keyboard.