Marcel Torne

Papers

Total Citations

H-Index

About

Marcel Torne is a rising researcher in robotics, whose work bridges the gap between imitation learning and reinforcement learning to achieve robust, precise manipulation. His key research areas include robot learning, sim-to-real transfer, and assembly tasks. Torne’s major contributions are twofold: first, he identified a critical limitation in Behavior Cloning (BC)—performance saturation in precision tasks—and proposed a solution in his 2025 paper “From Imitation to Refinement - Residual RL for Precise Assembly” (7 citations). This work introduces a residual reinforcement learning framework that refines BC policies, enabling robots to overcome the unreliability of pure imitation. Second, his 2024 paper “Reconciling Reality through Simulation: A Real-to-Sim-to-Real Approach for Robust Manipulation” (5 citations) tackles the challenge of policy robustness. By autonomously exploring in simulation, Torne’s method reduces the need for extensive human supervision while improving resilience to pose changes and disturbances. Though early in his career, his citation counts reflect growing recognition for addressing practical bottlenecks in robot learning. Torne’s work is notable for its direct impact on industrial assembly and real-world deployment, making him a promising voice in the quest for reliable, scalable robotic manipulation.