Logical Robotics is a consulting firm
specializing in deep learning for robotic applications
informed by human-robot interaction research
The foundation of robotic applications is skill development – using deep reinforcement learning, model-predictive control, and behavioral cloning. At Logical Robotics, we have years of experience building, testing, deploying, refining and benchmarking deep learned robotic navigation skills.
Large language model planning provides the foundation for translating human instructions into behavior and back into knowledge. Logical Robotics has experience with this cutting edge technology applied to visual-language navigation and question-answering problems.
Human-robot interaction research informs how robot skills and semantics are deployed in the service of helping people achieve their objectives. Logical Robotics is deeply connected to the social robot navigation community and is pushing for the development of social navigation standards.
We search for ways to create wonderful experiences for robotic applications without requiring massive computing resources beyond the reach of average applications. Logical Robotics is researching ways to make creating, training, tuning and deploying large models practical.