Camera-Based Traverse Control for Real-Time Autonomous Steering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing autonomous driving technologies face challenges in controlling precise movements in complex environments using nonlinear models, which are costly and computationally intensive, and often result in chattering or real-time control issues, especially when using expensive equipment like LiDAR.
Innovation Solution
A traverse direction control apparatus and method utilizing a camera sensor to calculate target point information and traverse direction control parameters based on a pre-stored kinematic model, enabling cost-effective and robust control using Single Input Single Output (SISO) methods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If nonlinear control methods are used for skid-steering mobile robots, then control precision is improved, but system complexity increases and real-time control becomes difficult
Solution Approach 1:
The control system is segmented into two independent parts: a nonlinear model for trajectory generation and a linear model for real-time control execution. This allows the nonlinear precision requirements to be satisfied in trajectory planning while the linear model handles real-time control with lower computational complexity.
Solution Approach 2:
The nonlinear model is used in advance to generate the desired trajectory and control commands. By performing the computationally intensive nonlinear calculations beforehand, the system avoids real-time nonlinear computation during actual control execution, enabling both precision and real-time performance.
2Measurement precision
If expensive equipment like LiDAR is used for autonomous driving, then measurement precision is improved, but cost increases
Solution Approach 1:
The system replaces expensive LiDAR sensors with inexpensive camera sensors. While cameras have limitations compared to LiDAR, the system compensates through sophisticated image processing algorithms and a dual-model control approach, achieving acceptable performance at much lower cost.
Solution Approach 2:
The system substitutes optical measurement (camera-based vision systems) for active electromagnetic sensing (LiDAR). By using camera sensors with computational imaging and processing, the system achieves functional equivalence or near-equivalence to LiDAR at a fraction of the cost.
3Manufacturing precision
If nonlinear models are used for robot control, then accuracy in complex environments is improved, but computational time increases
Solution Approach 1:
The control architecture segments computational tasks into offline trajectory generation using nonlinear models and online execution using linear models. This separation allows accurate trajectory planning without real-time computational burden, as the linear model can be executed rapidly during actual control.
Solution Approach 2:
All computationally intensive nonlinear calculations are performed in advance during trajectory generation. The pre-computed trajectory and control commands are then executed in real-time using simple linear model calculations, eliminating real-time nonlinear computation delays.
Data Source
AI summary
An embodiment of the present invention relates to a traverse direction control method, apparatus, and moving object for autonomous driving. More particularly, the traverse direction control apparatus for autonomous driving includes an information collection part for collecting data from one or more cameras mounted on an autonomous moving object; a target point information calculation module for calculating target point information on the basis of information collected from a camera module; and a traverse direction control parameter calculation module configured to obtain, based on the target point information, a traverse direction distance from the moving object at time t to a target point, an angle between a head of the moving object and the target point, and a present yaw rate of the moving object, and calculate a traverse direction control parameter of time t+1 based on a pre-stored kinematic model.