Topics include state estimation, planning in information space, partially observable Markov decision processes, reinforcement learning and planning with uncertain models.
Team MIT used the opportunity to showcase the power of our relatively formal approaches to perception, estimation, planning, and control.
We propose algorithms that find very good estimation plans efficiently, and discuss properties of synopsis structures that make them amenable to efficient use in an optimizer.
Estimation, planning, control, and learning are giving us robots that can generate good behavior given a specified objective and set of constraints.
Indeed, agile navigation through unknown, indoor environments poses a number of challenges for robotics research in terms of perception, state estimation, planning, and control.
In this paper, we identify and formalize the estimation planning problem in the context of a System R-style optimizer extended with complex synopsis techniques such as multi-dimensional histograms.
The review is organized around critical problems/issues including engineering, cost estimation, planning, scheduling, and monitoring and control of project operations, with the objectives to optimize cost and time through the efficient uses of constrained or unconstrained resources.
This leads to an "estimation planning" problem that has been ignored to date in literature: how to choose the most efficiently-computable collection of intermediate synopses to construct, so that all the required cardinality estimates can be computed as accurately as possible from the base-table synopses?
We first had a requirements analysis, then rapid prototyping and re design, budget estimation, business plan etc etc.
