Paper defines solver-sufficient symbols for hybrid robot planning
Original title: What Makes a Good Symbol for Hybrid Discrete-Continuous Planning?
A counterexample-guided method invents symbols that cut solving cost on simulated and real robots while keeping solution quality.
The paper argues that a useful planning symbol should expose discrete commitments that simplify the original continuous problem. It characterizes solver-sufficient symbolization for finite-horizon piecewise-affine systems and gives a counterexample-guided invention algorithm, then extends the idea to neural-network energy functions. Experiments on simulated and real-robot tasks report lower solving cost with preserved solution quality, plus better interpretability and steerability.
- Solver sufficiency is precisely characterized for finite-horizon piecewise-affine systems.
- Symbols can be induced from continuous optimization landscapes of neural energy functions.
- The authors say the theory also explains why some human-defined symbols help solvers.
From the original article
1 min read at huggingface.co
The physical world has continuous state and action spaces, yet robot planning often benefits from hybrid discrete-continuous formulations that explicitly expose combinatorial structure, such as contact modes or discrete primitive actions. Existing methods typically rely on human-defined symbol sets or symbols derived from language supervision, without asking what makes a symbol computationally useful.
By Yiyuan Pan, Changliu Liu, Jiayuan Mao · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.