An electronic apparatus is provided. The electronic apparatus includes a communicator comprising communication circuitry, a memory storing information on an artificial intelligence model, and a processor configured to: obtain a map generated based on sensing data obtained by an external electronic apparatus, simulate driving of the external electronic apparatus on the obtained map based on a plurality of parameter values and obtain driving result data for the plurality of parameter values, train the artificial intelligence model based on the plurality of parameter values and the obtained driving result data and obtain a plurality of parameter values related to driving of the external electronic apparatus, and control the communicator to transmit the plurality of obtained parameter values to the external electronic apparatus.

A system and method for implementing a neural network based vehicle dynamics model are disclosed. A particular embodiment includes: training a machine learning system with a training dataset corresponding to a desired autonomous vehicle simulation environment; receiving vehicle control command data and vehicle status data, the vehicle control command data not including vehicle component types or characteristics of a specific vehicle; by use of the trained machine learning system, the vehicle control command data, and vehicle status data, generating simulated vehicle dynamics data including predicted vehicle acceleration data; providing the simulated vehicle dynamics data to an autonomous vehicle simulation system implementing the autonomous vehicle simulation environment; and using data produced by the autonomous vehicle simulation system to modify the vehicle status data for a subsequent iteration.

A method of controlling a vehicle using a first model trained according to optimized output of a second model.

The invention proposes an automatic driving “machine consciousness” model, which is composed by the human's safety driving rules. Establish the dynamic fuzzy event probability measure relation, or fuzzy relation, or probability relation of the automatic driving vehicle and the surrounding passing vehicle. The decision result of “machine consciousness” of automatic driving vehicle is realized by complicated logic operation and using the antagonistic result of logic operation in both positive and negative directions. The implementation result is that it can make the decision-making result of automatic driving vehicle close to the result of human's biological consciousness, which can improve the safety of automatic driving vehicle, reduce the development cost and reduce the distance of road test.

Techniques for determining a vehicle action and controlling a vehicle to perform the vehicle action for navigating the vehicle in an environment can include determining a vehicle action, such as a lane change action, for a vehicle to perform in an environment. The vehicle can detect, based at least in part on sensor data, an object associated with a target lane associated with the lane change action sensor data. In some instances, the vehicle may determine attribute data associated with the object and input the attribute data to a machine-learned model that can output a yield score. Based on such a yield score, the vehicle may determine whether it is safe to perform the lane change action.

Systems, methods, tangible non-transitory computer-readable media, and devices associated with object perception and prediction of object motion are provided. For example, a plurality of temporal instance representations can be generated. Each temporal instance representation can be associated with differences in the appearance and motion of objects over past time intervals. Past paths and candidate paths of a set of objects can be determined based on the temporal instance representations and current detections of objects. Predicted paths of the set of objects using a machine-learned model trained that uses the past paths and candidate paths to determine the predicted paths. Past path data that includes information associated with the predicted paths can be generated for each object of the set of objects respectively.

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 information—e.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.

In one embodiment, a method by a computing system of a vehicle includes determining an environment of the vehicle. The method includes generating, based on the environment, multiple proposed vehicle actions with associated operational data. The method includes determining a comfort level for each proposed vehicle action by processing the environment and operational data using a model for predicting comfort levels of vehicle actions. The model is trained using records of performed vehicle actions. The record for each performed vehicle action includes environment data, operational data, and a perceived passenger comfort level for the performed vehicle action. The method includes selecting a vehicle action from the proposed vehicle actions based on the determined comfort level. The method includes causing the vehicle to perform the selected vehicle action.

An AI apparatus and an operating method are provided, the AI apparatus includes a communication interface to receive 3D sensor data and bumper sensor data from a first cleaner, a processor to generate surrounding situation map data based on the 3D sensor data and the bumper sensor data, and a learning processor to generate learning data by labeling area classification data for representing whether the surrounding situation map data corresponds to the stuck area, and to train a stuck area classification model based on the learning data. The processor transmits the trained stuck area classification model to a second cleaner through the communication interface.

A method includes generating a first simulated environment. The first simulated environment includes a route for a simulated vehicle. The method includes determining a set of locations within the first simulated environment for a set of objects. The method includes determining a path for the simulated vehicle based on the route and set of locations. The method includes generating a set of simulated environments based on the first simulated environment and set of locations. The method includes generating a set of images based on the set of simulated environments, the path, and the set of non-deterministically generated objects. The non-deterministically generated objects include unrealistic objects and optionally include realistic objects. The method includes training vehicle guidance models using the set of images which may include abstract or unrealistic objects. The trained vehicle guidance models may be directly used on real vehicles in corresponding real world environments.