Object Recognition Device Using Acceleration-Based Time Lag Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object recognition systems for ego vehicles face challenges in accurately estimating the state of detected objects when the ego vehicle is moving, as they struggle to distinguish between time lag and velocity errors, leading to incorrect estimation of time lag amounts and relative velocities.
Innovation Solution
An object recognition device and method that generate and output updated object data and corrected ego vehicle data by associating and predicting data from sensors, using a time measuring unit, data receiving unit, prediction processing unit, correlation processing unit, and update processing unit to correct ego vehicle data and predict object data, thereby improving accuracy in object state estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the time lag amount is estimated using the relative velocity of stationary objects and ego vehicle velocity, then the time lag can be corrected when the ego vehicle is moving, but the estimation fails when the ego vehicle is stopped or moving at constant velocity because both velocities remain unchanged
Solution Approach 1:
The patent changes the parameter used for time lag estimation from velocity-based (relative velocity of stationary objects) to acceleration-based (acceleration of the ego vehicle). By using acceleration information, the system can estimate time lag amounts even when the ego vehicle is stopped or moving at constant velocity, since acceleration remains detectable during motion changes. This resolves the contradiction by making the estimation method adaptable to all vehicle motion conditions while maintaining accuracy.
2Ease of manufacture
If the time lag amount is estimated from relative velocity data, then processing can be performed using available sensor data, but the error and time lag cannot be distinguished leading to erroneous estimation
Solution Approach 1:
The patent extracts acceleration information as a separate parameter from the velocity-based estimation approach. By using acceleration data independently, the system can distinguish between actual time lag effects and velocity measurement errors, since acceleration provides a different temporal perspective on the same physical phenomena. This separation allows for more accurate time lag estimation while maintaining relatively simple processing using standard sensor data.
3Loss of information
If multiple sensors are used to detect object data and ego vehicle data at different times, then comprehensive information can be gathered, but time synchronization issues arise requiring complex correction procedures
Solution Approach 1:
The patent applies preliminary action by using acceleration data to predict the ego vehicle velocity at the specific time when object data is detected. Instead of waiting to synchronize data after collection, the system proactively corrects the velocity timestamp using acceleration-based predictions. This preliminary correction simplifies the overall synchronization process while maintaining data completeness from multiple sensors operating at different times.
Data Source
AI summary
Provided are an object recognition device and the like configured to associate, for detected object data and detected ego vehicle data that are received from respective sensors in a period from a previous processing time to a current processing time, an object data time with each piece of the detected object data, and an ego vehicle data time with each piece of the detected ego vehicle data, predict the detected ego vehicle data at the object data time to generate a result of the prediction as corrected ego vehicle data, and predict updated object data at the object data time to generate a result of the prediction as predicted object data.