In multi-modal sensing setting, another challenging problem is how to sync the RGB sensory signals with other sensors (e.g. other RGB sensors, depth sensors, IMU, tactile sensors, etc.) This is the foundation of multi-modal perception. Yes, it is more engineering than “research”, let’s accumulate the experience and crush it!
Tactile sensing is a way of dense force sensing. It can meansure subtle force information, which can be used for many applications, such as human-robot interaction, object understanding, etc. To achieve high-density, sensitive, flexible tactile sensing is not easy. Current the cutting-edge solution is to adopt vision-gel-based design.
How to localize the agent is studied by the community of SLAM, but how to localize arbitrary objects? When you require high-precision localization, you may be surprised how little can current technology do.