Unmanned Vehicle Sensor Layout Using Simulation-Based Perception Tuning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sensor solutions for unmanned vehicles have weak perception capability and low perception accuracy, primarily due to reliance on physical parameters alone, which affects safe operation.
Innovation Solution
A method involving the establishment of a simulated unmanned vehicle and simulation scene for simulation driving, where sensor solutions are determined and corrected based on initialization parameters, considering physical parameters, obstacle shape, vehicle size, and traffic environment, using simulation data and perception algorithms to enhance perception accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If sensor solution is determined only based on physical parameters of the sensor, then device complexity is reduced, but perception capability and perception accuracy deteriorate
Solution Approach 1:
The patent applies preliminary action by conducting simulation experiments before final sensor solution deployment. A simulated unmanned vehicle with simulated sensors performs simulation driving in virtual scenes to pre-evaluate perception performance. This allows optimization of sensor parameters (installation positions, detection ranges, fields of view) before actual implementation, ensuring high perception accuracy while avoiding complex trial-and-error in real deployments.
Solution Approach 2:
The patent uses copying by creating a virtual copy of the unmanned vehicle and its sensors in a simulation environment. The simulated sensor solution replicates the physical sensor characteristics and installation configurations. By evaluating the copied virtual system's perception performance in simulated scenes, the patent optimizes the actual sensor solution without requiring multiple physical prototypes, thus reducing device complexity while maintaining high perception accuracy.
2Device complexity
If detection distance is estimated only based on obstacle size and vehicle model size, then measurement process is simplified, but detection distance accuracy deteriorates
Solution Approach 1:
The patent replaces the mechanical estimation method (based solely on physical dimensions) with a simulation-based computational approach. Instead of using simple geometric calculations from obstacle and vehicle sizes, the system uses simulated sensor data from virtual driving scenarios to empirically determine detection distances. This substitution of mechanical estimation with simulation-based measurement significantly improves detection distance accuracy while keeping the implementation feasible.
Solution Approach 2:
The patent applies parameter changes by transitioning from fixed geometric parameters (obstacle size, vehicle size) to dynamic simulation parameters (sensor detection ranges, fields of view, simulation scene configurations). By adjusting these parameters in the simulation environment based on actual performance requirements, the system achieves accurate detection distance estimation that adapts to different operating conditions, rather than relying on static dimensional calculations.
3Device complexity
If sensor solution does not consider obstacle shape and motion characteristics, then device complexity is reduced, but perception capability deteriorates
Solution Approach 1:
The patent applies dynamics by incorporating motion characteristics into the sensor solution design through simulation. Instead of static geometric considerations, the system simulates dynamic scenarios including moving obstacles with various shapes and motion patterns. The sensor parameters are optimized based on these dynamic simulation results, enabling the sensor solution to effectively track and perceive moving targets of different shapes and velocities, thus enhancing perception capability while maintaining reasonable design complexity.
Data Source
AI summary
The present application discloses a method, a device, equipment, and storage medium for determining a sensor solution, where the method includes: establishing a simulated unmanned vehicle and a simulation scene, where the simulation scene is used for the simulated unmanned vehicle to perform simulation driving; determining a first sensor solution according to a initialization parameter, and determining, according to the first sensor solution, simulation data generated by the simulated unmanned vehicle during the simulation driving in the simulation scene; and; and determining a first perception parameter of the first sensor solution according to the simulation data, and correcting the first sensor solution according to the first perception parameter to obtain a sensor solution applied to an unmanned vehicle.