Case Study: Heriot-Watt University accelerates robotics research with Google Cloud Platform

Heriot-Watt research team: Helping train robots to collaborate

Heriot-Watt University, based in Edinburgh, was working on robotics research that required training graph neural network and reinforcement learning models for open ad hoc teamwork. The team needed far more compute than local or campus resources could provide, especially for simulations running up to 10 million timesteps and multiple GPU instances. Google Cloud Platform, including Google Cloud Research Innovator support, provided the scalable infrastructure they needed.

With Google Cloud Platform, Heriot-Watt University could run eight GPUs simultaneously, complete training faster, and experiment more freely without scheduling campus servers. The university gained convenient on-demand access from anywhere, plus research credits and resources through the Research Innovator program, helping accelerate iteration and analysis for its robotics work.

Heriot-Watt University

Ignacio Carlucho

Assistant Professor, School of Engineering and Physical Sciences