IIIT-H wins Best Paper Award, presents 10 research papers at global robotics conference

HYDERABAD: Researchers from the International Institute of Information Technology Hyderabad (IIIT-H) presented 10 research papers and a workshop paper at the 2026 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), held in Pittsburgh, US. The institute also won the Best Paper Award at a conference workshop.
The award-winning paper, Beyond Frontiers: Scene-Anomaly Guided Autonomous Exploration, was presented at the workshop on Bridging Perspectives in Navigation: Lessons Learned and Future Directions. Authored by Akash Kumbar and Abhinav Raundhal under the supervision of Professor K Madhava Krishna, it proposes a framework called SCAGE that enables robots to identify poorly mapped areas and explore them to build more complete 3D representations of unfamiliar environments. The award carried a prize of $500 sponsored by Field AI.
The institute’s research covered robotic navigation, aerial manipulation, object grasping, perception, control and multi-drone mission planning.
Research focuses on safer, more capable autonomous robots
Among the featured studies, AERMANI-VLM explored how drones equipped with robotic arms could follow natural-language instructions while keeping decision-making separate from physical actions to improve safety.
Another paper, AeroGrab, presented a framework for helping drones identify and pick up objects in cluttered environments while avoiding collisions. Researchers also developed GlassFormer, which combines camera images with 60.5 GHz millimetre-wave radar to help robots detect transparent surfaces such as glass, particularly in poor lighting.
Other studies addressed real-time adaptation of aerial robots carrying different loads, contact detection without dedicated touch sensors, and more efficient robot navigation using pixel-level connections in visual maps.
Research on multi-drone inspection missions proposed a method to calculate spare fleet requirements, reporting 99.8% mission success across five test scenarios. Another project, SILICA, used knowledge from text-to-image AI models to detect glass and estimate its depth, while AeroPlace-Flow explored language-based object placement by aerial robots.
Workshop paper examines radar-based autonomous driving
IIIT-H researchers also presented Radar-Dot at the Long-Term Perception for Human-Centric Autonomy workshop. The study combines radar and camera images to estimate vehicle movement and identify moving objects, addressing challenges in autonomous driving.
Students from the institute’s Robotics Research Centre also chaired focused sessions on robot skill learning and vision-language robot agents.
Professor Madhava Krishna, head of the Robotics Research Centre, said the institute’s IROS results were significant because of the number of accepted papers and its 100% strike rate, compared with the conference’s overall acceptance rate of less than 37%.

