Brain Corporation | Date: 2014-04-29
A robotic vehicle may be operated by a learning controller comprising a trainable convolutional network configured to determine control signal based on sensory input. An input network layer may be configured to transfer sensory input into a hidden layer data using a filter convolution operation. Input layer may be configured to transfer sensory input into hidden layer data using a filter convolution. Output layer may convert hidden layer data to a predicted output using data segmentation and a fully connected array of efficacies. During training, efficacy of network connections may be adapted using a measure determined based on a target output provided by a trainer and an output predicted by the network. A combination of the predicted and the target output may be provided to the vehicle to execute a task. The network adaptation may be configured using an error back propagation method. The network may comprise an input reconstruction.
Brain Corporation | Date: 2014-11-14
A robotic device may be operated by a learning controller comprising a feature learning configured to determine control signal based on sensory input. An input may be analyzed in order to determine occurrence of one or more features. Features in the input may be associated with the control signal during online supervised training. During training, learning process may be adapted based on training input and the predicted output. A combination of the predicted and the target output may be provided to a robotic device to execute a task. Feature determination may comprise online adaptation of input, sparse encoding transformations. Computations related to learning process adaptation and feature detection may be performed on board by the robotic device in real time thereby enabling autonomous navigation by trained robots.
Brain Corporation | Date: 2014-12-31
A random k-nearest neighbors (RKNN) approach may be used for regression/classification model wherein the input includes the k closest training examples in the feature space. The RKNN process may utilize video images as input in order to predict motor command for controlling navigation of a robot. In some implementations of robotic vision based navigation, the input space may be highly dimensional and highly redundant. When visual inputs are augmented with data of another modality that is characterized by fewer dimensions (e.g., audio), the visual data may overwhelm lower-dimension data. The RKNN process may partition available data into subsets comprising a given number of samples from the lower-dimension data. Outputs associated with individual subsets may be combined (e.g., averaged). Selection of number of neighbors, subset size and/or number of subsets may be used to trade-off between speed and accuracy of the prediction.
Brain Corporation | Date: 2015-03-03
Apparatus and methods for detecting and utilizing saliency in digital images. In one implementation, salient objects may be detected based on analysis of pixel characteristics. Least frequently occurring pixel values may be deemed as salient. Pixel values in an image may be compared to a reference. Color distance may be determined based on a difference between reference color and pixel color. Individual image channels may be scaled when determining saliency in a multi-channel image. Areas of high saliency may be analyzed to determine object position, shape, and/or color. Multiple saliency maps may be additively or multiplicative combined in order to improve detection performance (e.g., reduce number of false positives). Methodologies described herein may enable robust tracking of objects utilizing fewer determination resources. Efficient implementation of the methods described below may allow them to be used for example on board a robot (or autonomous vehicle) or a mobile determining platform.
Brain Corporation | Date: 2014-03-13
Apparatus and methods for a modular robotic device with artificial intelligence that is receptive to training controls. In one implementation, modular robotic device architecture may be used to provide all or most high cost components in an autonomy module that is separate from the robotic body. The autonomy module may comprise controller, power, actuators that may be connected to controllable elements of the robotic body. The controller may position limbs of the toy in a target position. A user may utilize haptic training approach in order to enable the robotic toy to perform target action(s). Modular configuration of the disclosure enables users to replace one toy body (e.g., the bear) with another (e.g., a giraffe) while using hardware provided by the autonomy module. Modular architecture may enable users to purchase a single AM for use with multiple robotic bodies, thereby reducing the overall cost of ownership.