S. Gupta*, S. Sharma*, S. Gunagi, S. Sundaram
IROS 2026
Instead of giving each robot a smarter local controller, TOLD adapts who each robot listens to — rewiring edge weights online via gradient descent to minimize formation error. No training, no centralized coordinator, stackable on any existing method.
S. Sharma, A. Raizada, S. Sundaram
ArXiv
IRisPath fuses infrared and RGB camera streams to build robust costmaps for off-road robot navigation that remain reliable in both daylight and nighttime conditions without specialized calibration targets.
S. Gupta, S. Sharma, S. Sundaram
TMLR 2025 · 2025
A consensus-based decentralised cooperative monitoring system for agile targets using swarm robots under uncertainty accepted at Transactions on Machine Learning Research.
S. Sharma, V. HV, M. U. Kumari
IEEE INDICON 2023 · 2023
An automated litmus-test synthesis tool that generates distinguishing tests between two versions of a memory model, used for formal verification of ARM's memory architecture.
S. Sharma, N. A. Menezes
IEEE Connect 2022 · 2022
A priority-based decentralised algorithm for multi-robot path planning that resolves inter-agent conflicts without central coordination, demonstrated on a physical multi-robot testbed.