Vehicle Trajectory Planning Neural Network With External Waypoint Memory
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current neural network systems for autonomous and semi-autonomous vehicles lack efficient methods to plan trajectories that balance safety, comfort, and complexity, often relying on internal memory which can lead to memory loss and increased computational expense.
Innovation Solution
A computer-implemented neural network system that processes waypoint data, environmental data, and navigation data to generate optimal trajectory locations, utilizing an external memory to store previous waypoints and reduce system complexity, while incorporating a trajectory management system to select and update waypoints based on generated scores.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If internal memory is used to store previous waypoints in neural network systems, then trajectory planning can be performed, but memory loss and increased computational complexity occur
Solution Approach 1:
The system segments the memory function by separating short-term memory (internal neural network state for immediate processing) from long-term memory (external storage for historical waypoint data). This division allows the neural network to access historical trajectory information without burdening its internal memory, thereby reducing computational complexity while maintaining planning reliability.
Solution Approach 2:
An external memory module acts as an intermediary between the neural network and historical waypoint data. This mediator stores and retrieves long-term trajectory information, allowing the neural network to access past waypoints without internal memory loss, thus resolving the contradiction between reliable trajectory planning and computational complexity.
2Measurement precision
If multiple sensing channels (LIDAR, RADAR, camera) are integrated for real-time environmental data processing, then safety and accuracy improve, but system complexity and computational expense increase
Solution Approach 1:
The system merges multiple sensing channels (LIDAR, RADAR, camera) into a unified environmental perception framework. By integrating these diverse sensors and their data streams into a single coordinated system, the patent achieves comprehensive environmental accuracy while managing complexity through unified processing architecture.
Solution Approach 2:
The external memory system serves multiple functions: storing historical waypoints, caching environmental data from various sensors, and providing contextual information for trajectory planning. This multi-functionality reduces the need for separate specialized systems, thereby improving measurement precision without proportionally increasing system complexity.
3Adaptability or versatility
If neural networks process long sequences of trajectory data, then better generalization is achieved, but computational expense and memory requirements increase
Solution Approach 1:
The system transitions from processing long sequences in the temporal dimension to accessing historical data in the spatial dimension of external memory storage. By organizing trajectory data spatially in external memory rather than processing it sequentially in internal memory, the network achieves better generalization from long sequences without the corresponding increase in computational energy consumption.
Data Source
AI summary
Systems, methods, devices, and other techniques for training a trajectory planning neural network system to determine waypoints for trajectories of vehicles. A neural network training system can train the trajectory planning neural network system on the multiple training data sets. Each training data set can include: (i) a first training input that characterizes a set of waypoints that represent respective locations of a vehicle at each of a series of first time steps, (ii) a second training input that characterizes at least one of (a) environmental data that represents a current state of an environment of the vehicle or (b) navigation data that represents a planned navigation route for the vehicle, and (iii) a target output characterizing a waypoint that represents a target location of the vehicle at a second time step that follows the series of first time steps.