Convolutional Network for Robotic Vehicle Adaptive Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Remote control of robotic devices is inadequate when dynamics of the control system and environment change rapidly, requiring user attention and relying on user experience, which can lead to inadequate performance in handling unexpected obstacles.
Innovation Solution
A computerized neuron network with an input layer, intermediate layer, and output layer that processes sensory input to determine control signals, updates learning parameters based on performance measures, and adjusts trajectories to align closer with target trajectories, enabling adaptive control and obstacle avoidance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If remote control operation is used, then user can control robotic device from distance, but user attention and experience are required which becomes inadequate when dynamics change rapidly
Solution Approach 1:
The robotic device performs self-learning and self-adjustment through autonomous operation. The system executes tasks autonomously while learning from performance measures and feedback, gradually improving its own control capabilities without requiring continuous user intervention or attention.
Solution Approach 2:
The system implements a feedback mechanism where performance measures are determined based on evaluation of executed trajectories versus target trajectories. This feedback is used to update learning parameters, enabling the robotic device to continuously improve its performance and adapt to changing dynamics.
2Adaptability or versatility
If autonomous learning is implemented, then robotic device can adapt to changing environments, but system complexity increases with neural network architecture
Solution Approach 1:
The neural network is segmented into distinct functional layers: input layer for receiving sensory data, intermediate layer for processing and feature extraction, and output layer for generating control signals. This segmentation allows the complex learning task to be distributed across specialized sub-components, managing overall system complexity.
Solution Approach 2:
The system employs dynamic learning parameters that are continuously updated based on performance feedback. The learning parameters are not fixed but adapt over time as the robotic device learns from its experiences, enabling the system to maintain adaptability while managing complexity through progressive learning rather than pre-configured complexity.
Data Source
AI summary
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.