Brian Acosta

Papers

Total Citations

H-Index

About

Brian Acosta is a leading roboticist whose research bridges the critical gap between simulation and real-world bipedal locomotion. His work centers on three interconnected pillars: validating robotics simulators for real-world impact, developing robust control strategies for underactuated bipedal robots, and integrating vision-based planning for dynamic walking. Acosta’s most influential paper, “Validating Robotics Simulators on Real-World Impacts” (2022, 32 citations), addresses a fundamental gap in the field by systematically comparing simulation fidelity against physical robot performance—a cornerstone for researchers relying on sim-to-real transfer. His 2023 work on “Bipedal Walking on Constrained Footholds with MPC Footstep Control” (14 citations) tackles the challenging problem of agile, underactuated bipeds navigating discontinuous terrain, introducing model predictive control for footstep planning on small, weak-ankled platforms. Most recently, his 2025 paper on “Learning a Vision-Based Footstep Planner for Hierarchical Walking Control” (1 citation) pioneers a vision-driven approach that replaces fragile manual pipelines, enabling real-time footstep planning in unstructured environments. Acosta’s contributions are shaping the next generation of dynamic, perceptive bipedal robots, making him a rising authority in legged locomotion and simulation validation.