A method for training a control policy for a robot device includes acquiring a reference state of an environment of the robot device and a reference observation of the environment for the reference state. The method also includes generating, for each of a plurality of errors of an estimation of a pose of the robot device, an observation that is disturbed with respect to the reference observation according to the error of the pose estimation and a training data element comprising the generated observation as a training input. The method further includes training the control policy using the generated training data elements.

In conventional robot navigation techniques learning and planning algorithms act independently without guiding each other simultaneously. A method and system for robotic navigation with simultaneous local path planning and learning is disclosed. The method discloses an approach to learn and plan simultaneously by assisting each other and improve the overall system performance. The planner acts as an actuator and helps to balance exploration and exploitation in the learning algorithm. The synergy between dynamic window approach (DWA) as a planning algorithm and a disclosed Next best Q-learning (NBQ) as a learning algorithm offers an efficient local planning algorithm. Unlike the traditional Q-learning, dimension of Q-tree in the NBQ is dynamic and does not require to define a priori.

Systems and methods are described for reacting to a feature in an environment of a robot based on a classification of the feature. A system can detect the feature in the environment using a first sensor on the robot. For example, the system can detect the feature using a feature detection system based on sensor data from a camera. The system can detect a mover in the environment using a second sensor on the robot. For example, the system can detect the mover using a mover detection system based on sensor data from a lidar sensor. The system can fuse the data from detecting the feature and detecting the mover to produce fused data. The system can classify the feature based on the fused data and react to the feature based on classifying the feature.

A method and apparatus of generating a path for autonomous navigation of a mobile computing apparatus is provided. The mobile computing apparatus includes one or more processors configured to map a set global path, which connects a current position of the mobile computing apparatus with a future destination of the mobile computing apparatus based on a topology map, to a real-time generated occupancy grid map that represents, based on sensor data of a surrounding environment of the mobile computing apparatus, an un-drivable area and a drivable area in the surrounding environment, and generate a local path, for a traveling of the mobile computing apparatus, connecting the current position with a sub-waypoint that corresponds to the drivable area in response to a next waypoint, to which the mobile computing apparatus is set to travel based on the global path, being determined to correspond to the un-drivable area.

In order to achieve the purpose, a robotic cleaner according to an aspect of the present disclosure includes: a traveling unit for moving a body in a traveling region; a distance measuring sensor for acquiring distance sensing information about a distance to an object outside the body; and a control unit which generates a grid map about the traveling region from the distance sensing information, performs, when dividing the traveling region into a plurality of sub-areas, ray casting on a plurality of traveling nodes on a path of the grid map with respect to each sub-area to search for an open space, and sets an open node for the open space to calculate a topology graph between the traveling nodes and the open node. Therefore, efficiency can be improved by minimizing unnecessary traveling when traveling for searching for a space in which additional traveling is required. Furthermore, avoidance traveling can be reduced by setting the cleaner to travel along the center of a passage during additional search traveling.

A work machine includes a work implement. A control system for the work machine includes a processor. The processor acquires actual topography data indicative of an actual topography of a work site. The processor acquires work data including a width of a work implement. The processor generates work path data based on the actual topography data and the work data. The work path data indicates positions of a plurality of work paths aligned in a lateral direction. The processer determines a work order of the plurality of work paths based on the work path data. The processer controls a work machine to perform work according to the work paths in the work order.

An electronic device and a control method thereof are provided. The electronic device includes a communication interface, memory storing one or more computer programs, and one or more processors communicatively coupled to the communication interface and the memory. The one or more computer programs include computer-executable instructions that, when executed by the one or more processors, cause the electronic device to obtain a 2-dimensional (2D) map of an indoor space generated based on sensing data, and traveling data obtained by at least one external device, simplify the obtained 2D map, based on the traveling data, and generate a 3-dimensional (3D) map, based on the simplified 2D map.

A method for controlling at least one autonomous mobile robot, wherein the at least one robot is designed to navigate within a functional area on the basis of a map of said functional area, and to perform at least one task autonomously in the functional area. The method involves: receiving a job request which contains instructions for carrying out at least one task in the functional area, automatically dividing the job request into at least two sub-tasks, and automatically determining a sequence in which said sub-tasks are to be processed by the at least one robot, the job request being fully completed once all sub-tasks have been processed.

Method and system for navigating an autonomous vehicle in an open-pit site. The method involves acquiring multiple observations and odometry data from various poses while driving, accessing a topological map of the site with intersections and segments; accessing an observational map of the site with past observations including surroundings information linked to an intersection or segment, processing past observations, acquired observations, and odometry data, applying particle filtering techniques and Gaussian Processes to model observations from discrete poses as a continuous variable and estimate the current pose, statistically predicting the next pose based on the current direction of movement; commanding the autonomous vehicle via actuators controlled by the processor unit, based on detecting whether the vehicle is in a segment or intersection, and issuing either a moving-forward instruction to traverse the segment or a steering instruction to take a subsequent segment.

Provided are a method, system and device for global path planning for an unmanned vehicle in an off-road environment. The method includes: obtaining satellite elevation data and a satellite remote sensing image of a current off-road environment; constructing a digital elevation model (DEM); determining slope and land surface relief of each grid in the current off-road environment; performing gray processing on the satellite remote sensing image to obtain grayscale values of the grids; determining traversal costs of the grids corresponding to different ground types; constructing a global grid map based on the slope and the land surface relief of each grid, as well as the traversal costs corresponding to the different ground types; determining a rugged terrain potential field and path costs; and searching for paths using a Bresenham's line algorithm and Theta* algorithm based on the rugged terrain potential field and the path costs, to generate a global path.