SoftVTBench targets deformable-object manipulation
The dataset pairs visuo-tactile observations with FEM ground truth to measure deformation-aware success.
Why it matters
The work addresses a gap in robotics evaluation: policies can complete deformable-object tasks while slipping or over-compressing objects. Adding physical-interaction quality to benchmarks could push models toward safer, more controlled manipulation.
The key points
- 1.SoftVTBench includes 4,000 expert demonstrations.
- 2.The benchmark evaluates task success and deformation tolerance.
- 3.Data covers deformable objects and visually matched rigid twins.
Researchers introduced SoftVTBench, a visuo-tactile dataset and benchmark for deformable-object manipulation. The dataset includes 4,000 expert demonstrations and more than 50 assets, with synchronized multi-view RGB, tactile RGB and marker motion, proprioception, language, gripper actions, and evaluator-only finite-element states at 20 Hz. The benchmark defines a Deformation-aware Success Rate that requires task completion while keeping peak normalized deformation within tolerance.
⚡ Try this today
Use DSR-style deformation limits when evaluating policies for soft or deformable-object handling.
Sources & original reporting
This brief summarizes and links to reporting from the publishers below.
Enjoyed this brief? Get the next one in your inbox.
More in Research
Startup says AI is not close to curing cancer
TechCrunch reports the company argues better data is the missing piece.
HarnessRisk benchmark tests agent harness safety
The benchmark evaluates safety failures across agent harness lifecycle phases.
Agent Lightning v1.0 targets harnessed agentic RL
The framework connects arbitrary agent harnesses to RL training through an LLM endpoint proxy.