Pete Senior

About

Pete Senior is a researcher pushing the boundaries of reinforcement learning (RL) for robotics, with a focus on bridging the critical gap between simulation and real-world hardware. His primary research areas include hardware-software co-optimisation and robust, generalisable control policies. Senior’s most notable contribution is the ORCHID framework, which enables the simultaneous optimisation of a robot’s physical hardware design and its control policy during RL training—a significant departure from traditional methods that treat hardware as immutable. This work, his most cited with 9 citations, addresses a fundamental bottleneck in robotic design. He further tackles the challenge of brittle convergence and sim-to-real transfer with HARL-A (Hardware Agnostic Reinforcement Learning Through Adversarial Selection), a method that uses adversarial training to improve policy generalisation across unseen environments. By directly confronting issues of data scarcity and overfitting, Senior’s research is laying the groundwork for more adaptable and resilient robotic systems, moving beyond narrow, simulation-specific solutions toward truly hardware-agnostic intelligence.

Research Focus

Key Achievements

Top Papers

1. 1
   ORCHID: Optimisation of Robotic Control and Hardware In Design using Reinforcement Learning

   9 citations · 2021
2. 2
   HARL-A: Hardware Agnostic Reinforcement Learning Through Adversarial Selection

   2 citations · 2021

Key Collaborators

Contact & Links