Subjective Localization with Action Respecting Embedding

Abstract

Summary. Robot localization is the problem of how to estimate a robot's pose within an objective frame of reference. Traditional localization requires knowledge of two key conditional probabilities: the motion and sensor models. These models depend critically on the specific robot as well as its environment. Building these models can be time-consuming, manually intensive, and can require expert intuitions. However, the models are necessary for the robot to relate its own subjective view of sensors and motors to the robot's objective pose. In this paper we seek to remove the need for human provided models. We introduce a technique for subjective localization, relaxing the requirement that the robot localize within a global frame of reference. Using an algorithm for action-respecting non-linear dimensionality reduction, we learn a subjective representation of pose from a stream of actions and sensations. We then extract from the data natural motion and sensor models defined for this new representation. Monte Carlo localization is used to track this representation of the robot's pose while executing new actions and receiving new sensor readings. We evaluate the technique in a synthetic image manipulation domain and with a mobile robot using vision and laser sensors.

Efficient, generalized indoor WiFi GraphSLAM

“…The location of the nodes in GraphSLAM are in a different reference frame than those in the ground-truth trajectory. To report error in standard units, as in [1], we compute localization accuracy with the subjective-objective technique of [33]: We characterize "how accurately a person detects returning to a previously visited location". For each time t i during our trajectory, we have an inferred 'subjective' location x ti from GraphSLAM and a true 'objective' location x * ti for the same timestamp in the ground truth.…”

Efficient 3D Scene Modeling and Mosaicing

“…Initial proposals on image similarity cross-over detection use image representations based on a single global descriptor, embodying visual content such as color or texture [13,70,72,117,141]. Such global descriptors are sensitive to camera viewpoint and illumination changes, decreasing the robustness of the cross-over detection.…”

Learning GP-BayesFilters via Gaussian process latent variable models

“…Furthermore, the tracking application introduced by Urtasun and colleagues [19] is not designed for real-time or near real-time performance, nor does is provide uncertainty estimates as GP-BayesFilters. Other alternatives for non-linear embedding in the context of dynamical systems are hierarchical GPLVMs [12] and action respecting embeddings (ARE) [1]. None of these techniques are able to incorporate control information or impose prior knowledge on the structure of the latent space.…”