Today, the mobility and manipulation capabilities of robots are tightly coupled to the hardware of the system. Since robot architectures are fixed from the conception of the design, both physical and functional capabilities of each robot are limited to its physical architecture. Soft robots address this limitation of conventional robots and we regard robotic origamis (Robogamis) as soft robots that could transform their body shape and functionality to interact and adapt to the unknown environment. In this talk, I will address the specific challenges involved with the building of Robogamis and Reconfigurable Robotics Lab's research goal to provide robust solutions to meet the engineering challenges needed to demonstrate a capable end-to-end Robogami system that starts with specifications and delivers a fully functional robot.
The number of robotics applications in which we need SLAM as traditionally understood is depressingly small. What we really need is vast scale, robust, change-tolerant localisation. Sure, we need to build the maps first and they need to be special and plastic to model change. We can tolerate them being expensive to build but they must be cheap to use. I'll talk about some of the work we are doing in the Oxford Mobile Robotics group borne from this view which might enable low cost, large scale mobile robot (and car) navigation over large scales and long periods. And I won't mention SLAM.