A Differentiable Distance Metric for Robotics Through Generalized Alternating Projection
Vinicius M. Gonçalves, Shiqing Wei, Eduardo Malacarne S. de Souza, Anthony Tzes, and Farshad Khorrami · 2025
Read A Differentiable Distance Metric for Robotics Through Generalized Alternating Projection by Vinicius M. Gonçalves, Shiqing Wei, Eduardo Malacarne S. de Souza, Anthony Tzes, and Farshad Khorrami (2025) 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-2507-01181.txt.
Description
Vinicius Gonçalves and co-authors build a distance measure for robots that stays smooth enough for control math, in a 2025 paper. Controllers need not only distance to obstacles but also its rate of change. The plain geometric distance has kinks that break derivatives. A recent smooth alternative had practical drawbacks. Their version rests on a generalized alternating projection method.
Simpler formulas give the smooth projection for box-shaped obstacles. Unlike the earlier metric, theirs falls to zero when objects overlap. That property matters for safety functions that must trigger on contact. Experiments show the measure working in practice. Code ships inside a public Python simulation package. Barrier-based controllers can now use the derivative safely. The authors aim at everyday robotics, where convex shapes dominate collision checks.
Catalog record
| CREATOR | Vinicius M. Gonçalves, Shiqing Wei, Eduardo Malacarne S. de Souza, Anthony Tzes, and Farshad Khorrami |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 45 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2507-01181