Search for: All records

Localization of mobile robots is essential for navigation and data collection. This work presents an optical localization scheme for mobile robots during the robot’s continuous movement, despite that only one bearing angle can be captured at a time. In particular, this paper significantly improves upon our previous works where the robot has to pause its movement in order to acquire the two bearing angle measurements needed for position determination. The latter restriction forces the robot to work in a stop-and-go mode, which constrains the robot’s mobilitty. The proposed scheme exploits the velocity prediction from Kalman filtering, to properly correlate two consecutive measurements of bearing angles with respect to the base nodes (beacons) to produce location measurement. The proposed solution is evaluated in simulation and its advantage is demonstrated through the comparison with the traditional approach where the two consecutive angle measurements are directly used to compute the location.

 

The problem of localizing a moving target arises in various forms in wireless sensor networks. Deploying multiple sensing receivers and using the time-difference-of-arrival (TDOA) of the target’s emitted signal is widely considered an effective localization technique. Traditionally, TDOA-based algorithms adopt a centralized approach where all measurements are sent to a predefined reference node for position estimation. More recently, distributed TDOA-based localization algorithms have been shown to improve the robustness of these estimates. For target models governed by highly stochastic processes, the method of nonlinear filtering and state estimation must be carefully considered. In this work, a distributed TDOA-based particle filter algorithm is proposed for localizing a moving target modeled by a discrete-time correlated random walk (DCRW). We present a method for using data collected by the particle filter to estimate the unknown probability distributions of the target’s movement model, and then apply the distribution estimates to recursively update the particle filter’s propagation model. The performance of the distributed approach is evaluated through numerical simulation, and we show the benefit of using a particle filter with online model learning by comparing it with the non-adaptive approach.

 

We propose Deterministic Sequencing of Exploration and Exploitation (DSEE) algorithm with interleaving exploration and exploitation epochs for model-based RL problems that aim to simultaneously learn the system model, i.e., a Markov decision process (MDP), and the associated optimal policy. During exploration, DSEE explores the environment and updates the estimates for expected reward and transition probabilities. During exploitation, the latest estimates of the expected reward and transition probabilities are used to obtain a robust policy with high probability. We design the lengths of the exploration and exploitation epochs such that the cumulative regret grows as a sub-linear function of time. 

We study the problem of multi-robot coverage over an unknown, nonuniform sensory field. Modeling the sensory field as a realization of a Gaussian Process and using Bayesian techniques, we devise a policy that aims to balance the tradeoff between learning the sensory function and covering the environment. We propose an adaptive coverage algorithm called Deterministic Sequencing of Learning and Coverage (DSLC) that schedules learning and coverage epochs such that its emphasis gradually shifts from exploration to exploitation while never fully ceasing to learn. Using a novel definition of coverage regret which characterizes the overall coverage performance of a multi-robot team over a time horizon T, we analyze DSLC to provide an upper bound on expected cumulative coverage regret. Finally, we illustrate the empirical performance of the algorithm through simulations of the coverage task over an unknown distribution of wildfires. 

We consider a scenario in which an autonomous vehicle equipped with a downward-facing camera operates in a 3D environment and is tasked with searching for an unknown number of stationary targets on the 2D floor of the environment. The key challenge is to minimize the search time while ensuring a high detection accuracy. We model the sensing field using a multi-fidelity Gaussian process that systematically describes the sensing information available at different altitudes from the floor. Based on the sensing model, we design a novel algorithm called Expedited Multi-Target Search (EMTS) that (i) addresses the coverage-accuracy trade-off: sampling at locations farther from the floor provides a wider field of view but less accurate measurements, (ii) computes an occupancy map of the floor within a prescribed accuracy and quickly eliminates unoccupied regions from the search space, and (iii) travels efficiently to collect the required samples for target detection. We rigorously analyze the algorithm and establish formal guarantees on the target detection accuracy and the detection time. We illustrate the algorithm using a simulated multi-target search scenario.