Real-time navigation in crowded dynamic environments using Gaussian process motion control

Abstract

In this paper, we propose a novel Gaussian process motion controller that can navigate through a crowded dynamic environment. The proposed motion controller predicts future trajectories of pedestrians using an autoregressive Gaussian process motion model (AR-GPMM) from the partiallyobservable egocentric view of a robot and controls a robot using an autoregressive Gaussian process motion controller (AR-GPMC) based on predicted pedestrian trajectories. The performance of the proposed method is extensively evaluated in simulation and validated experimentally using a Pioneer 3DX mobile robot with a Microsoft Kinect sensor. In particular, the proposed method shows over 68% improvement on the collision rate compared to a reactive planner and vector field histogram (VFH).

Chance-constrained target tracking for mobile robots

“…Figure 1 shows an illustration, in which a mobile robot has detected a target from time k − 2 to k − 1 ( Figure 1(a)) and moves to a new location to make sure the target is within robot's sensing range (Figure 1(b)). We assume that the distribution of the next position of the target is available using a motion prediction algorithm, such as Kalman filters or the autoregressive Gaussian process motion model [14]. Let p(k) = [x T (k) y T (k)] T be the position of the target at time k. From the motion prediction algorithm, the target's position at time k using measurements up to time k − 1 has the Gaussian distribution with meanp(k) and covariance Σ T (k).…”

Augmented dictionary learning for motion prediction

“…(9)). Then, we perform a line search from the previous dictionary D old to the global optimum D * to ensure that the update step would decrease the objective value 2 . In practice, we interweave the two update steps to achieve faster convergence and still attain theoretical convergence properties.…”

“…With wide application of onboard sensors and GPS devices, large volumes of pedestrian and vehicular trajectory data has been collected [1], [2]. These datasets are useful for understanding the mobility patterns of human and vehicles, which in turn, benefit applications such as autonomous navigation [2] and mobility-on-demand systems [3].…”

“…These datasets are useful for understanding the mobility patterns of human and vehicles, which in turn, benefit applications such as autonomous navigation [2] and mobility-on-demand systems [3]. To ensure safety in these applications, it is important that the autonomous vehicles can accurately anticipate the future paths of their surrounding moving objects.…”

Vehicle navigation in populated areas using predictive control with environmental uncertainty handling

“…However coexisting with humans and operating efficiently in such environment, requires a robot must be able to navigate in harmony with traffic participants -humans. Thus the problem of automated navigation in dynamic environments has become an important challenge of contemporary Robotics [1] [3] [5] [8] [9]. In contrast to static and supervised environments, navigating robot in dynamic and uncertain conditions requires many issues to be solved e.g.…”