Priyanka Mandikal

Indian Institute of Science Bangalore, The University of Texas at Austin

Papers

Total Citations

H-Index

About

Priyanka Mandikal is a researcher whose work spans robotics, computer vision, and human motion modeling, with a particular focus on advancing the capabilities of intelligent systems in understanding and replicating complex human behaviors. Her most prominent contributions lie in the domain of dexterous robotic manipulation, where she has pioneered methods for teaching multi-fingered robotic hands to grasp objects with human-like agility. Her landmark work on object-centric visual affordances embeds rich perceptual representations within deep reinforcement learning frameworks, enabling robots to overcome the formidable challenges posed by high-degree-of-freedom manipulation — work that has garnered over 16 citations across multiple publications. Her DexVIP framework further advances this agenda by leveraging in-the-wild human-object interaction videos as priors for robot learning, bridging the gap between human dexterity and robotic capability. Beyond robotics, Mandikal has contributed to human motion synthesis through cross-conditioned recurrent networks capable of generating realistic long-term inter-person interactions, earning 21 citations and finding relevance in animation, surveillance, and human-robot interaction. Her research consistently demonstrates a commitment to grounding robotic learning in human-inspired representations and real-world data.