What is multi-modal sensor fusion, and when would you use it for Berkshire Grey's perception pipeline?
