Qi Lu
University of Texas Rio Grande Valley, TX·Computer Science
Picture a team of robots sent to search a large, unknown area for people after a disaster, to harvest crops across a farm, or to explore another planet. This kind of task is called foraging: finding and collecting targets spread over a wide space.
- Fields
Computer Science
Engineering
- Topics
robot swarms
swarm robotics
heterogeneous robots
- Grants
- 1 known grant, through Sep 2027
What they work on
- Develop a decentralized algorithm that plans multiple shortest routes at once, so that UAVs (aerial drones) can explore promising locations on demand, collecting data quickly while saving energy for the ground robots in the swarm.
- Design decentralized federated learning algorithms, which let robots train a shared model without pooling their raw data, so that each drone or ground robot gets a customized, privacy-preserving model as the swarm grows.
- Build an early anomaly detection module based on sub-trajectory discords (unusual stretches of a robot's path) to catch robot failures or attacks quickly with limited computing power, before they spread through the swarm.
More about the research
Swarms made up of a single type of ground robot can work efficiently, but limited sensing and mobility slow them down when targets are scattered over large areas. Three problems stand in the way. Ground-only swarms have trouble covering large areas. The machine learning models that control robot behavior are usually trained on one central server, which does not scale well to many different kinds of robots. And a broken sensor or a cyberattack on one robot can spread failures through the whole swarm. This project proposes to tackle all three by combining robotics, machine learning and cybersecurity in a mixed swarm of UAVs (unmanned aerial vehicles, or drones) and UGVs (unmanned ground vehicles).
Why it matters
More dependable robot swarms could help with search and rescue, agricultural harvesting and space exploration, where large unknown areas must be covered quickly. Teams of drones and ground robots that are energy-efficient, protect data privacy and can detect sensor failures or attacks early could make these systems more practical and safer to deploy in real field conditions.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Jiawei Yuan
University of Massachusetts-Dartmouth, MA·Computer Science
Design and build an integrated, extensible research platform for securely combining AI with unmanned aerial vehicles (drones). The platform is meant to support studies that find weaknesses and test defenses.
Miao Pan
University of Houston, TX·Computer Science
Design software-hardware co-designed model pruning and adaptive quantization schemes that exploit the sparsity of model updates to reduce computation and memory access during federated learning on a single 5G mobile device.
Joseph Camp
Southern Methodist University, TX·Computer Science
Develop an AI curriculum covering federated learning, explainable AI, homomorphic encryption and neural networks, taught through lectures, exercises and simulations and shared via the NAIRR Pilot.
The school
University of Texas Rio Grande Valley: 2 labs on PI Pages, $91M a year on research, 50 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.