PISTO: Proximal Inference for Stochastic Trajectory Optimization
Hongzhe Yu, Zinuo Chang, Yongxin Chen · 2026
Read PISTO: Proximal Inference for Stochastic Trajectory Optimization by Hongzhe Yu, Zinuo Chang, Yongxin Chen (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-07215.txt.
Description
Hongzhe Yu, Zinuo Chang, and Yongxin Chen stabilize a popular robot planning method by uncovering its hidden structure. STOMP plans trajectories by sampling, handling costs that gradients cannot touch. The authors show it quietly minimizes a divergence from a probability distribution over trajectories, giving it a variational inference reading.
Their PISTO algorithm adds a penalty that keeps each update near the previous proposal, a trust-region style safeguard with simple closed-form updates. Sampling estimates those updates without derivatives. On robot arm benchmarks, PISTO succeeded far more often than two rival planners, with shorter, smoother paths. It also won on contact-rich locomotion and manipulation tasks.
Catalog record
| CREATOR | Hongzhe Yu, Zinuo Chang, Yongxin Chen |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 31 KB |
| LICENSE | http://creativecommons.org/publicdomain/zero/1.0/ |
| RIGHTS | CC0 |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.07215