James Motes

I am currently on the 2025–26 faculty job market in Computer Science, ECE, and Robotics. For faculty candidate content, please visit my Personal Webpage.

During my tenure as a postdoc (and prior to that as a PhD student) with the Parasol Lab at the University of Illinois Urbana-Champaign, my research has focused on advancing multi-robot task and motion planning (MR-TMP) methodologies to enhance the efficiency and adaptability of autonomous systems. I have worked with large teams of graduate (and undergraduate students) along with several industrial partners on robotics applications such as smart manufacturing and biofabrication labs in addition to some non-robotics applications including 3D routing and other engineering design tools.

While there have been many projects over the years (I joined the lab as an undergraduate in 2018 at Texas A&M before it moved to UIUC in 2019), there are three frameworks which have emerged which I am the most excited about moving forward: DaSH, ARC, and SPITE.

Decomposable state Space Hypergraph (DaSH): This multi-robot task and motion planning framework, the culmination of my PhD dissertation work, achieved up to three orders of magnitude speed up in planning time over competitor methods in multi-manipulator rearrangement planning problems. It leverages hypergraphs to model varying (de)compositions of the planning space to focus coordination (and the necessary computational effort) only where the problem demands, leveraging cheaper, faster decoupled planning when possible. For more details, see the publication.

In the initial paper, we successfully planned for 20 objects and 4 manipulators in a single scene. In our ongoing extensions, we have successfully planned for 60 objects with 16 manipulators (look for this soon from Scott Lee). Additionally, we have leveraged the framework for multi-robot motion planning in congested environments, extending prior work with Courtney McBeth to now handle problems with up to 128 robots and account for kinodynamic constraints.

Adaptable Robot Coordination (ARC): This multi-robot motion planning framework, similar to DaSH, attempts to focus coordination for multi-robot problems only where it is needed, and perform cheaper, decoupled planning whenever possible. Where DaSH uses the hypergraph structure to reason over where the coordination is needed in task and motion planning problems, ARC dynamically adjusts the coordination required during motion planning as projected conflicts are discovered, creating local subproblems and applying a hierarchy of increasing coordinated (and expensive) methods to resolve the conflict. This is can be incorporated into the larger DaSH framework, but it is impactful anywhere multi-robot motion planning is performed. For more details, see the publication.

We recently have extended the framework to account for kinodynamic constraints (K-ARC) and to leverage a database of previous conflict resolution experiences (E-ARC), further extending the applications and effectiveness of the framework.

Simple Polyhedral Intersection Techniques for modified Environments (SPITE): This framework leverages both over and under approximations of the geometry of robots and moveable objects in an environment to rapidly recompute validity of motions in the context of object movement. This is done by preprocessing a configuration space roadmap, generating approximate geometries of the swept volume of edges which can then be quickly validated against the moved geometry of an obstacle. For more details, see the publication.

Initial experiments considered discrete changes to the environment, where a new instance of the problem is loaded with changes to obstacle positions. This is both useful when considering a robot returning to a scene after other agents have been active, or when considering the implications of object rearrangement in task and motion planning problems (the focus of our soon to appear ICRA workshop paper).

We are actively extending the approach to function in the presence of actively moving obstacles, enabling real-time updates of robot motion plans in response to object movement.

Papers with Parasol Lab:

Updated: