Prasoon Goyal
Papers
4
Total Citations
26
H-Index
3
About
Prasoon Goyal is a leading researcher at the intersection of reinforcement learning (RL), natural language processing, and embodied AI. His work focuses on enabling robots and agents to understand and act upon complex, real-world environments using language as a guiding signal. Goyal’s most cited work, "PixL2R: Guiding Reinforcement Learning Using Natural Language by Mapping Pixels to Rewards" (2020, 13 citations), tackles the critical challenge of sparse rewards in RL by translating natural language instructions directly into reward functions, drastically reducing the number of interactions needed for an agent to learn. He further advanced the field by introducing "CH-MARL: A Multimodal Benchmark for Cooperative, Heterogeneous Multi-Agent Reinforcement Learning" (2022, 6 citations), a pioneering vision-and-language benchmark that simulates collaborative tasks among diverse robots in a home environment. Demonstrating a commitment to practical, helpful AI, Goyal also led the creation of the "Don't Forget to Put the Milk Back!" dataset (2024, 5 citations), which equips embodied agents with the ability to detect dangerous or unsanitary anomalies in the home, such as a forgotten stove or spilled milk. His contributions to the Alexa Prize SimBot Challenge further underscore his impact on building conversational, embodied agents for the real world.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4