BayesContact treats robotic perception as Bayesian inference. A particle-based belief represents uncertainty over object pose. Visual observations provide an initial estimate, while subsequent contact observations eliminate pose hypotheses that are inconsistent with physical interaction. The framework also studies how a robot should select informative contact actions. **Role in my research:** inference **Topics:** Bayesian inference · active sensing · pose estimation · contact [GitHub](https://github.com/kamiradi/bayes-contact) 