James D. Motes

Postdoc building adaptive planning systems for robot teams.

James D. Motes is a postdoctoral researcher in the Parasol Lab at the University of Illinois Urbana-Champaign, working with Nancy M. Amato, and he is a 2026-27 faculty candidate in computer science, ECE, robotics, and related areas.

James develops adaptive planning systems for robot teams: systems that decompose large multi-robot problems, coordinate only when tasks, conflicts, or workspace structure require it, and accelerate the motion validation bottlenecks that limit deployment. His work asks when robot teams should coordinate, how that coordination can run fast enough to use, and how people can specify, inspect, and revise robot behavior through language, visualization, shared task representations, and human-facing planning interfaces.

This page summarizes Parasol research projects he has helped lead in multi-robot planning, task and motion planning, adaptive coordination, motion planning acceleration, and human-steerable robot systems. For faculty candidate materials, see James’s personal website.

Quick links: CV / Google Scholar / GitHub / Email

Research Overview

In Parasol, we develop planning frameworks for robot teams that must operate in shared, cluttered, and changing environments. Our work studies how to decompose large multi-robot planning problems, guide computation toward the interactions that require coordination, and accelerate motion validation and roadmap update primitives so planning systems can respond quickly enough for real deployments.

Across these projects, the shared principle is that robot teams should not pay the cost of full coordination everywhere. Some methods predict structure from tasks or workspaces, some react to conflicts by creating local subproblems, some accelerate validation and roadmap updates, and some expose planning representations to people so they can inspect and guide robot behavior.

Research Thread Map

Thread

Decompose

We represent large planning problems through smaller task, robot, object, or workspace structures so robot teams avoid full joint space reasoning unless the problem demands it.

DaSH MR-TAMP Workspace-DaSH
Explore thread
Thread

Coordinate

We develop methods that identify where robots must reason together, resolve conflicts locally, and return to cheaper decoupled planning when possible.

ARC CIPHER Multi-Agent Systems
Explore thread
Thread

Guide

We use task structure, workspace topology, region decompositions, and other planning signals to focus computation on the regions and interactions that matter.

Workspace Guidance CIPHER Topological Guidance
Explore thread
Thread

Accelerate

We build planning infrastructure that reduces the cost of collision checking, roadmap validation, repeated queries, and replanning in changed or dynamic environments.

SPITE Dynamic Roadmaps Validation
Explore thread
Thread

Interact

We connect planning systems to human-facing interfaces so people can inspect, steer, and understand robot behavior in richer planning environments.

ERUPT XR Planning Interactive Planning
Explore thread
Thread

Deploy

We connect planning methods to larger Parasol applications in robot teams, manipulation, manufacturing, automated labs, biofabrication, routing, and engineering design tools.

Manufacturing Robot Teams Design Tools
Explore thread

Project Areas

Task & Motion Planning

DaSH Family

DaSH object rearrangement scenario

DaSH uses hypergraphs to represent how robot, object, and composite planning spaces split and merge during multi-robot task and motion planning.

Lazy-DaSH adds hierarchical task planning and lazy motion validation, while Workspace Guided-DaSH connects decomposition to congested multi-robot motion planning.

Hypergraph Planning MR-TAMP Lazy Validation
View project
Adaptive Coordination

ARC Family

ARC warehouse planning scenario

ARC begins with cheaper individual robot planning, detects inter-robot conflicts, and constructs local subproblems around interactions that require coordination.

Conflict Resolution Hybrid Planning Kinodynamics
View project
Workspace Guidance

CIPHER and Guided Coordination

CIPHER workspace decomposition

CIPHER uses workspace decompositions to guide multi-robot motion planning, refine coordination regions, and avoid full joint space planning when smaller searches suffice.

CIPHER MAPF Guidance Workspace Refinement
View project
Planning Infrastructure

SPITE and Fast Validation

SPITE dynamic obstacle environment

SPITE updates motion planning roadmaps when objects or obstacles move, classifying roadmap edges as valid, invalid, or uncertain before exact collision checking is needed.

Dynamic Roadmaps Collision Checking Open-SPITE
View project
Human-Facing Planning

ERUPT and Interactive Planning

ERUPT extended reality planning interface

ERUPT supports interactive robot motion planning in immersive virtual or augmented reality environments, connecting XR interaction with robotics middleware and planning tools.

Extended Reality MoveIt Interactive Planning
View project
Multi-Agent Systems

MAPF and Team-Level Planning

Multi-agent systems planning overview

Multi-agent planning provides the team-level coordination layer behind many Parasol robot-team systems. This work studies how robots share constraints, allocate tasks, avoid path conflicts, and coordinate routes before lower-level motion planners refine executable trajectories.

MAPF, CBS extensions, decentralized planning, and related multi-agent methods help connect discrete coordination with continuous motion planning, task and motion planning, and deployment-oriented robot teams.

MAPF CBS Decentralized Planning
View area
Applications

Deployment Contexts

Collaborative industrial robots

Manufacturing, logistics, automated labs, biofabrication, construction, and engineering design applications motivate the need for decomposable, adaptive, fast, and inspectable robot-team planning.

Manufacturing Automated Labs Engineering Design
View area

How the Work Connects

DaSH predicts coordination needs from task and planning-space structure. ARC reacts to conflicts by creating local subproblems. CIPHER and workspace-guided methods use environment structure to guide coordination. SPITE and related acceleration work reduce the cost of roadmap validation, collision checking, and replanning. Human-facing planning projects expose robot behavior and planning scenes to people so they can inspect, guide, and revise plans. Together, this work studies how scalable planning systems can become faster, more adaptive, and more usable in real environments.

Collaborate and Learn More

Students and collaborators interested in adaptive robot-team planning, task and motion planning, human-guided autonomy, accelerated motion planning, robot-team systems, manufacturing, automated labs, engineering design, or field robotics can explore the linked project pages and Parasol research areas, then contact James or the relevant Parasol project leads.

Useful entry points:

Papers with Parasol Lab:

Updated: