Navigating unmapped environments
There has been significant progress made when it comes to robots perceiving and navigating their environments. Just look at self-driving cars, for example. Mapping and navigation techniques will continue to evolve, but future robots need to be able to operate in environments that are unmapped and poorly understood.
Some of the improvement that need to be made include:
- How to learn, forget, and associate memories of scenes both qualitatively and semantically
- How to surpass purely geometric maps to have semantic understanding of the scene
- How to reason about new concepts and their semantic representations and discover new objects or classes in the environment through learning and active interactions
“For navigation, the grand challenge is to handle failures and being able to adapt, learn, and recover. For exploration, it is developing the innate abilities to make and recognize new discoveries,” the study says. “From a system perspective, this requires the physical robustness to withstand harsh, changeable environments, rough handling, and complex manipulation. The robots need to have significant levels of autonomy leading to complex self-monitoring, self-reconfiguration, and repair such that there is no single point of complete failure but rather graceful system degradation. When possible, solutions need to involve control of multiple heterogeneous robots; adaptively coordinate, interface, and use multiple assets; and share information from multiple data sources of variable reliability and accuracy.”
Well yeah. Most of the self-driving cars are designed in such a way so that they'll be able to provide safe driving for people in city or crowded areas. The objective is to somehow have an understanding of following traffic laws and clearing traffic. Of course this is something which is pretty much trending these days, but the points which are taken into consideration is how to use these cars in an urban or a dirt track which is filled with obstacles. Its not easy since the geometric maps of a city based areas are technically the same means learning from the environment in its future use is something which might prove a little difficult for self driving cars the way they are, that's one reason why these cars are still in development to run in different environment and course.