Itamar Mishani

Planning long-horizon manipulation is challenging because it requires reasoning over both which skills to employ and where they are likely to succeed. We present MOSAIC, a skill-centric framework that treats skills as active stakeholders in the search, using a high-fidelity physics simulator to evaluate their feasibility during planning. MOSAIC employs two complementary skill families: Generators, which identify "islands of competence" where skills are demonstrably effective, and Connectors, which link these skill-trajectories by solving boundary value problems. This removes the need for hand-coded skill preconditions and effects. We show that MOSAIC consistently achieves high success rates while reducing planning time compared to existing approaches.

This paper presents PACHS, a parallel best-first search algorithm for inference that combines both components of actor-critic architecture. The actor network generates actions, while the critic network provides cost-to-go estimates to guide the search. The approach employs dual parallelism through batch processing by neural networks and multi-threaded graph expansion.

This paper introduces Multi-Graph Search (MGS), a novel search-based motion planning approach designed for high-dimensional robotic systems like manipulators and mobile manipulators. The algorithm generalizes traditional unidirectional and bidirectional search to work with multiple implicit graphs simultaneously. It maintains and incrementally expands multiple implicit graphs over the state space, focusing exploration on high-potential regions while allowing disconnected subgraphs to merge through feasible transitions.

This paper introduces the Behavioral Constant-Time Motion Planner (B-CTMP), which extends Constant-Time Motion Planning (CTMP) to two-step manipulation tasks: a collision-free motion to a behavior initiation state, followed by execution of a manipulation behavior such as grasping or insertion. Since such behaviors are increasingly stochastic (e.g., learned skills), B-CTMP builds neighborhoods in object-pose space and caches a plan only if repeated rollouts statistically certify its success rate above a user-specified threshold, with guarantees that hold across the entire cache at a prescribed confidence level. Online queries return certified plans, or reject infeasible object poses, within a fixed time budget of milliseconds. We evaluate B-CTMP on shelf picking, plug insertion, and wheel replacement in simulation and on real robots.

Diffusion models have recently been successfully applied to a wide range of robotics applications for learning complex multi-modal behaviors from data. However, prior works have mostly been confined to single-robot and small-scale environments due to the high sample complexity of learning multi-robot diffusion models. In this paper, we propose a method for generating collision-free multi-robot trajectories that conform to underlying data distributions while using only single-robot data. Our algorithm, Multi-robot Multi-model planning Diffusion (MMD), does so by combining learned diffusion models with classical search-based techniques – generating data-driven motions under collision constraints. Scaling further, we show how to compose multiple diffusion models to plan in large environments where a single diffusion model fails to generalize well. We demonstrate the effectiveness of our approach in planning for dozens of robots in a variety of simulated scenarios motivated by logistics environments.

An exciting frontier in robotic manipulation is the use of multiple arms at once. However, planning concurrent motions is a challenging task using current methods. The high-dimensional composite state space renders many well-known motion planning algorithms intractable. Recently, Multi-Agent Path-Finding (MAPF) algorithms have shown promise in discrete 2D domains, providing rigorous guarantees. However, widely used conflict-based methods in MAPF assume an efficient single-agent motion planner. This poses challenges in adapting them to manipulation cases where this assumption does not hold, due to the high dimensionality of configuration spaces and the computational bottlenecks associated with collision checking. To this end, we propose an approach for accelerating conflict-based search algorithms by leveraging their repetitive and incremental nature – making them tractable for use in complex scenarios involving multi-arm coordination in obstacle-laden environments. We show that our method preserves completeness and bounded sub-optimality guarantees, and demonstrate its practical efficacy through a set of experiments with up to 10 robotic arms.