To enable highly accurate person identification processing by executing person identification processing in a case where it is confirmed that an illumination state at the time of capturing an image is similar to an illumination state at the time of capturing a learning image from which collation feature information has been acquired. A person identification unit that inputs a camera-captured image, acquires feature information for person identification from the input image, and executes collation processing between the acquired feature information and the collation feature information to execute the person identification processing is included. The person identification unit determines whether or not the illumination state at the time of capturing the input image is similar to the illumination state at the time of capturing the learning image from which the collation feature information has been acquired with reference to an illumination control program generated in advance, and executes the person identification processing to which the input image is applied in a case where the similarity is determined.

Systems, methods, tangible non-transitory computer-readable media, and devices for autonomous vehicle operation are provided. For example, a computing system can receive object data that includes portions of sensor data. The computing system can determine, in a first stage of a multiple stage classification using hardware components, one or more first stage characteristics of the portions of sensor data based on a first machine-learned model. In a second stage of the multiple stage classification, the computing system can determine second stage characteristics of the portions of sensor data based on a second machine-learned model. The computing system can generate an object output based on the first stage characteristics and the second stage characteristics. The object output can include indications associated with detection of objects in the portions of sensor data.

An unmanned device for a marine environment comprises a location sensor configured to gather location data corresponding to the unmanned device; at least one propulsion system; a transmitter and memory including computer program code. The computer program code is configured to, when executed, cause the processor to cause the propulsion system to propel the unmanned device in a pattern along the body of water, cause the sonar transducer to emit the one or more sonar beams into the body of water, receive sonar return data corresponding to sonar returns, and generate a sonar image corresponding to the sonar return data. Further, the computer program code is configured to cause the processor to detect an object within the sonar image, assign a score to the object indicating the likelihood that the object is a desired object type, and send an alert to the remote electronics device upon assignment of the score.

In a moving robot and a control method according to the present disclosure, an obstacle is detected using structured light irradiated in a predetermined type of light pattern in a traveling direction while traveling, and a specified operation is performed in response to the obstacle. Moreover, a dangerous obstacle is recognized by extracting changes over time using a plurality of images for an obstacle or a low obstacle that is difficult to determine as detected data, and thus, it is possible to improve accuracy according to the determination of the obstacle, improve a corresponding operation according to the obstacle, minimize the uncleaned area while preventing restraint due to the obstacle, and improve the cleaning performance.

In various examples, one or more output channels of a deep neural network (DNN) may be used to determine assignments of obstacles to paths. To increase the accuracy of the DNN, the input to the DNN may include an input image, one or more representations of path locations, and/or one or more representations of obstacle locations. The system may thus repurpose previously computed informatione.g., obstacle locations, path locations, etc.from other operations of the system, and use them to generate more detailed inputs for the DNN to increase accuracy of the obstacle to path assignments. Once the output channels are computed using the DNN, computed bounding shapes for the objects may be compared to the outputs to determine the path assignments for each object. Additionally, a machine may perform control operations based at least on the path assignments.

Disclosed is an intelligent learning and adjustment system for a tennis training robot, including an image recognition system, an algorithm model, a back-end processing platform, and an optimization model. Preprocessing of incoming ball data is performed, various necessary data, such as speeds and directions of flying tennis balls, spinning and placements, are collected, various data sets are processed by using various machine learning algorithms, effective predictions and decisions are generated to facilitate the prediction of the placement and difficulty level of the incoming ball, so that a capability and level of a sparring athlete can be evaluated, the tennis training robot accordingly makes prediction and recognition, and carries out interactive feedback actions in a timely manner. The entire training process involves continuously updating of weights and bias values to make the predictions increasingly accurate, and the tennis training robot can provide an interactive intelligent training method.

Methods, systems, and apparatus, including computer programs encoded on computer-storage media, for robot navigation. In some implementations, a method includes obtaining sensor data captured by one or more sensors located at a property over a time period; detecting an object represented in the sensor data; detecting, using the detected object and multiple subsets of the sensor data, a movement pattern of the object over the time period; determining an area navigable for a robot at the property using the detected movement pattern of the object over the time period; and providing, to the robot, an indication of the area navigable for the robot.

A notification control apparatus (2000) acquires video data or audio data for a plurality of persons (10) (a human group (40)) who have a conversation within a predetermined distance with each other in a surveillance area. The notification control apparatus (2000) determines whether or not the human group (40) is in a state suitable for receiving a notification using the acquired data. When the human group (40) is in the state suitable for receiving the notification, the notification control apparatus (2000) issues a predetermined notification.

A system for faster object attribute and/or intent classification may include an machine-learned (ML) architecture that processes temporal sensor data (e.g., multiple instances of sensor data received at different times) and includes a cache in an intermediate layer of the ML architecture. The ML architecture may be capable of classifying an object's intent to enter a roadway, idling near a roadway, or active crossing of a roadway. The ML architecture may additionally or alternatively classify indicator states, such as indications to turn, stop, or the like. Other attributes and/or intentions are discussed herein.

In various examples, one or more output channels of a deep neural network (DNN) may be used to determine assignments of obstacles to paths. To increase the accuracy of the DNN, the input to the DNN may include an input image, one or more representations of path locations, and/or one or more representations of obstacle locations. The system may thus repurpose previously computed informatione.g., obstacle locations, path locations, etc.from other operations of the system, and use them to generate more detailed inputs for the DNN to increase accuracy of the obstacle to path assignments. Once the output channels are computed using the DNN, computed bounding shapes for the objects may be compared to the outputs to determine the path assignments for each object.