Adaptive Bayes Filter Attenuation for Road Profile Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing road profile estimation methods for autonomous vehicles suffer from inertia in Bayes filters, leading to delayed adaptation of road profiles and uneven driving behavior, particularly when detecting curves or obstacles late.
Innovation Solution
Adapting the attenuation of Bayes filters based on situational factors, such as vehicle speed and distance, by adjusting parameters like process noise in the Kalman filter to balance responsiveness and stability, and using uncertainty analysis from surroundings measurement data to dynamically regulate the estimation process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If the attenuation of the Bayes filter is kept constant to maintain stable lateral guidance, then the driving behavior remains even and stable, but the reaction speed to detect curves or obstacles is delayed
Solution Approach 1:
The attenuation parameter of the Bayes filter is transformed from a static constant value to a dynamic variable that adapts based on the vehicle's operating conditions. The controller adjusts the attenuation in real-time according to vehicle speed and distance to obstacles, enabling the system to respond quickly when needed while maintaining stability during normal operation.
Solution Approach 2:
The invention changes the parameter values of the Bayes filter dynamically. Specifically, the attenuation parameter is modified based on vehicle speed and distance measurements, allowing the filter to switch between high-attenuation mode (for stability) and low-attenuation mode (for fast response) depending on the situation.
2Loss of time
If the attenuation is decreased to increase reaction speed for late curve detection, then the adaptation of the road profile becomes faster, but the lateral guidance becomes uneven and unstable
Solution Approach 1:
The system dynamically adjusts the attenuation parameter based on real-time vehicle speed and distance measurements. When the vehicle is moving fast or approaching a curve, the attenuation is decreased to enable faster adaptation. When the vehicle is moving slowly on straight sections, the attenuation is increased to maintain smooth lateral guidance.
Solution Approach 2:
Different attenuation values are applied locally based on the specific driving situation. The controller evaluates the current context (speed, distance, curve detection status) and applies the appropriate attenuation level locally, rather than using a uniform attenuation value throughout all operating conditions.
3Stability of the object's composition
If the Bayes filter uses high attenuation to stabilize road profile estimation, then the lateral guidance becomes smoother, but the system cannot adapt quickly to newly detected curves or obstacles
Solution Approach 1:
The attenuation parameter is made dynamic and context-dependent. The controller monitors vehicle speed and distance to obstacles, and adjusts the attenuation accordingly. This allows the system to maintain high attenuation (for stability) during normal cruising, and switch to low attenuation (for adaptability) when curves or obstacles are detected.
Solution Approach 2:
The invention implements parameter changes in the Bayes filter by modifying the attenuation value based on operating conditions. The system transitions between different parameter states (high attenuation for stability, low attenuation for adaptability) depending on the vehicle's speed and distance to potential hazards.
Data Source
AI summary
Systems, methods, and apparatuses are provided for estimating a road profile in surroundings of a vehicle by using a state function that describes the road profile and a Bayes filter for adapting the state function based on surroundings measurement data. Sensor-captured surroundings measurement data is received. Information about an uncertainty regarding at least either the state function or the sensor-captured surroundings measurement data is received. The information is taken as a basis for adapting a state inaccuracy that influences an attenuation of the Bayes filter.