{"id":"arxiv/2607.24959","url":"https://arxiv.org/abs/2607.24959","tags":["cs.RO","cs.DC","cs.LG","eess.SY","math.OC"],"score":0,"title":"Amortising Trajectory Optimisation for Residual MPC via Implicit Contact Differentiation","domain":"cs.RO","source":"arxiv","authors":["Daniel Layeghi","Thomas Corbères","Calum Arnott","Aditya Kamireddypalli","Hashim Al-Obaidi","Steve Tonneau","Michael Mistry"],"reviews":0,"subject":"cs.RO","summary":"Differentiable simulation can accelerate contact-rich trajectory optimisation by exposing local sensitivities of task outcomes to controls. Existing approaches either use finite differences, which are expensive and step-size sensitive; differentiate iterative contact solvers by unrolling automatic differentiation (AD), which stores a growing computation trace; or require intricate, solver-specific KKT sensitivity derivations. We introduce an AD-assisted implicit derivative for regularised smooth contacts and apply it to Mujoco MJX, based on the Implicit Function Theorem (IFT). The method…","category":"research","comments":0,"disputed":false,"division":"cs","mentions":0,"discussed":0,"published":"2026-07-27","reactions":{}}