For a mobile robot, navigation skills that are safe, efficient, and socially compliant in crowded, dynamic environments are essential. This is a particularly challenging problem as it requires the robot to accurately predict pedestrians’ movements, analyse developing traffic situations, and plan its own path or trajectory accordingly. Previous approaches still exhibit low accuracy for pedestrian trajectory prediction, and they are prone to generate infeasible trajectories under complex crowded conditions. In this paper, we develop an improved socially conscious model to learn and predict a pedestrian’s future trajectory. To generate more efficient and safer trajectories in a changing crowed space, an online path planning algorithm considering pedestrians’ predicted movements and the feasibility of the candidate trajectories is proposed. Then, multiple traffic states are defined to guide the robot finding the optimal navigation strategies under changing traffic situations in a crowded area. We have demonstrated the performance of our approach outperforms state-of-the-art approaches with public datasets, in low-density and simulated medium-density crowded scenarios.

For a service robot, learning appropriate behaviours to acquire task knowledge and deliberation in various situations is essential, but the existing methods do not support merging the plan-based activity experiences from multiple situations in the same task. In this paper, an abstract method is introduced to integrate the empirical activity schemas of multiple situations, and a novel algorithm is presented to learn activity schema with abstract methods. Furthermore, a novel planner called the schema-based optimized planner (SBOP) is developed based on the learned activity schema, in which actions merging optimization and partially backtracking techniques are adopted. A simulation with a PR2 robot and a physical experiment is conducted to validate the proposed method. The results show that the robot can generate a plan to recover from failure automatically using the novel learning and planning method, given that the experienced exception has been merged in the activity schema. Owing to the improved autonomy, the proposed SBOP exhibits increased efficiency in dealing with tasks containing loops and multiple activity schema instances. This research presents a novel solution of how to merge activity experiences from multiple situations and generate an intelligent and efficient plan that could adapt to a dynamically changing environment.

Industrial landscape (IL) is the spatial consequence of industry behaviors. Aboundance of industrial spatial pattern and dependence can be observed due to the variation of different geographical factors. Location of work-force and capital also differ along the earth surface. Because of data collection difficulty, administrative area is often used as the basic spatial analysis unit and high spatial resolution less introduced. In the paper, data of the 2nd National Economic Sensus of China are employed to carry out high spatial resolution IL analysis. Spatial statistics methods are conducted to find industry cluster, special industry zone and spatial dependence at 1Km² resolution grid in GIS software. It's argued that subtle spatial segmentation will violate industry characters in some spatial cells in which only few enterprise sites. Therefore supernumerary filters should be introduced to exclude these cells.