Two-Stage Focused Inference for Resource-Constrained Collision-Free Navigation

Abstract

Abstract-Long-term operations of resource-constrained robots typically require hard decisions be made about which data to process and/or retain. The question then arises of how to choose which data is most useful to keep to achieve the task at hand. As spacial scale grows, the size of the map will grow without bound, and as temporal scale grows, the number of measurements will grow without bound. In this work, we present the first known approach to tackle both of these issues.The approach has two stages. First, a subset of the variables (focused variables) is selected that are most useful for a particular task. Second, a task-agnostic and principled method (focused inference) is proposed to select a subset of the measurements that maximizes the information over the focused variables. The approach is then applied to the specific task of robot navigation in an obstacle-laden environment. A landmark selection method is proposed to minimize the probability of collision and then select the set of measurements that best localizes those landmarks. It is shown that the two-stage approach outperforms both only selecting measurement and only selecting landmarks in terms of minimizing the probability of collision. The performance improvement is validated through detailed simulation and real experiments on a Pioneer robot.

Attention and Anticipation in Fast Visual-Inertial Navigation

“…The problem looks for a subset of features S, containing no more than κ features, which optimizes the task performance f (·). This is a standard feature selection problem and has been used across multiple fields, including machine learning [69], robotics [39], and sensor networks [43]. Problem (1) is NPhard [46] in general.…”

Attention and anticipation in fast visual-inertial navigation

“…The contributions that are most relevant to our proposal are the one of Davison [37], Lerner et al [38], Mu et al [39], Wu et al [40], and Zhang and Vela [41]. The pioneering work of Davison [37] is one of the first papers to use information theoretic constructs to reason about visual features, and shares many of the motivations discussed in the present paper.…”

“…The problem looks for a subset of features S, containing no more than κ features, which optimizes the task performance f (•). This is a standard feature selection problem and has been used across multiple fields, including machine learning [69], robotics [39], and sensor networks [43]. Problem (1) is NPhard [46] in general.…”

No belief propagation required: Belief space planning in high-dimensional state spaces via factor graphs, the matrix determinant lemma, and re-use of calculation

“…BSP, in both augmented and non-augmented cases, is another important problem, where in contrast to the former case, only a subset of variables is of interest (see e.g. Krause et al, 2008; Levine and How, 2013; Mu et al, 2015). For example, one can look for an action that reduces the uncertainty of the robot’s final pose.…”

“…For example, one can look for an action that reduces the uncertainty of the robot’s final pose. The complexity of such a problem is much higher and proposed techniques succeeded to solve it in O ( k n 3 ) (Krause et al, 2008; Levine and How, 2013) with k being the size of candidate actions set, and in O ( true n ~ 4 ) (Mu et al, 2015) with true n ~ being the size of involved clique within a Markov random field representing the system.…”