Ananye Agarwal

I am a founding research scientist at Skild AI. I did my PhD at Carnegie Mellon University with Deepak Pathak. I am interested in building general-purpose, omni-bodied robot brains that work across tasks, environments, and robots.

I graduated from IIT Delhi with a B.Tech. in Computer Science and a President's Gold Medal (equivalent to valedictorian). In the past, I have been lucky to work with Prof. Mausam on neuro-symbolic AI and with Prof. Manik Varma on Extreme Classification. I have also had the good fortune of interning at Microsoft Research India where I worked with Dr. Ankit Garg on algebraic complexity. I also have a gold medal in the International Physics Olympiad (IPhO) in 2017.

Recent work at Skild AI

• Generalist locomotion via in-context learning
  [Chainsaw vs. Robot] [loss of limbs] [blog]
• End-to-end perceptive locomotion
  [video] [blog]
• Humanoid parkour
  [full video] [Skild AI]
• Multi-agent soccer via self-play
  [full video] [blog]

Press

Publications

• LocoFormer: Generalist Locomotion via Long-Context Adaptation
  Min Liu, Deepak Pathak, Ananye Agarwal
  CoRL 2025, Best Paper Award Finalist [webpage] [paper] [arXiv] [video] [blog]
• 
  SAPG: Split and Aggregate Policy Gradients
  Jayesh Singla*, Ananye Agarwal*, Deepak Pathak
  ICML 2024, Oral [webpage]
• SPIN: Simultaneous Perception, Interaction and Navigation
  Shagun Uppal, Ananye Agarwal, Haoyu Xiong, Kenneth Shaw, Deepak Pathak
  CVPR 2024 [webpage] [arXiv]
• Dexterous Functional Grasping
  Ananye Agarwal, Shagun Uppal, Kenneth Shaw, Deepak Pathak
  CoRL 2023 [webpage] [arXiv]
• Extreme Parkour with Legged Robots
  Xuxin Cheng*, Kexin Shi*, Ananye Agarwal, Deepak Pathak
  ICRA 2024 [webpage] [arXiv] [code]
• LEAP Hand: Low-Cost, Efficient, and Anthropomorphic Hand for Robot Learning
  Kenneth Shaw, Ananye Agarwal, Deepak Pathak
  RSS 2023 [webpage] [paper]
• Legged Locomotion in Challenging Terrains using Egocentric Vision
  Ananye Agarwal*, Ashish Kumar*, Jitendra Malik^†, Deepak Pathak^†
  CoRL 2022, Best Systems Paper Award [webpage] [arXiv] [demo] [in the media]
• Coupling Vision and Proprioception for Navigation of Legged Robots
  Zipeng Fu*, Ashish Kumar*, Ananye Agarwal, Haozhi Qi, Jitendra Malik, Deepak Pathak
  CVPR 2022, Best Paper at Multimodal Learning Workshop [webpage] [pdf] [arXiv] [code] [video]
• 
  End-to-End Neuro-Symbolic Architecture for Image-to-Image Reasoning Tasks
  Ananye Agarwal, Pradeep Shenoy, Mausam
  [arXiv]
• 
  SiameseXML: Siamese networks meet extreme classifiers with 100M labels
  K. Dahiya, A. Agarwal, D. Saini, K. Gururaj, J. Jiao, A. Singh, S. Agarwal, P. Kar and M. Varma
  ICML 2021 [PDF] [code]

Live demos

Our small low-cost robot can perceive and traverse challenging terrain. This uses a single neural network running onboard that directly maps pixels to joint torques.

CoRL 2021, Auckland

CVPR 2021, New Orleans