Cooperative Bicycle Trajectory Prediction for Obstructed Collision Mitigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicular safety systems are ineffective in mitigating collisions between motorized vehicles and bicycles due to limited visibility, unpredictable bicycle trajectories, and complex environmental factors, leading to increased collision risks.
Innovation Solution
A vehicle collision mitigation system that utilizes onboard sensors and cooperative vehicle-to-vehicle communication to obtain and process environmental and bicycle information, predict trajectories, and provide feedback to drivers to avoid collisions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If onboard sensors are used to detect bicycles, then collision risk can be mitigated, but detection effectiveness deteriorates when line of sight is obstructed by other vehicles or objects
Solution Approach 1:
The system combines multiple sensing modalities (radar, camera, LIDAR) to create a comprehensive detection system. Radar provides all-weather, obstruction-penetrating detection while cameras provide visual confirmation, and LIDAR adds depth information. This multi-sensor fusion enables reliable bicycle detection even when line of sight is partially obstructed by other vehicles or objects.
Solution Approach 2:
The system introduces an intermediary processing layer that fuses data from multiple sensors and uses machine learning algorithms to infer bicycle presence and trajectory even when direct detection is obscured. The intermediary system combines partial observations from different angles and sensors to reconstruct complete bicycle state information.
2Device complexity
If traditional safety systems are used, then system simplicity is maintained, but effectiveness against bicycles deteriorates due to limited visibility and unpredictable trajectories
Solution Approach 1:
The system implements dynamic trajectory prediction that continuously adapts to changing bicycle motion patterns. The machine learning model updates predictions in real-time based on observed bicycle behavior, environmental conditions, and sensor data, enabling the system to handle unpredictable trajectories while maintaining manageable complexity through incremental learning.
Solution Approach 2:
The system incorporates feedback loops where prediction results are continuously validated against new sensor observations. When discrepancies are detected, the system adjusts its prediction models and re-evaluates collision risk. This feedback mechanism improves reliability without requiring complete system redesign, maintaining a balance between complexity and effectiveness.
3Difficulty of detecting and measuring
If cooperative V2V communication is implemented, then detection capability in obstructed conditions improves, but system complexity and communication requirements increase
Solution Approach 1:
The system uses universal V2V communication protocols that enable multiple functions through a single communication infrastructure. The same communication channel is used for sharing bicycle detection data, trajectory predictions, and collision risk information between vehicles. This multi-functional approach reduces overall system complexity compared to implementing separate dedicated systems for each function.
4Measurement precision
If machine learning models are used to predict bicycle trajectories, then prediction accuracy improves, but computational requirements and processing time increase
Solution Approach 1:
The system performs preliminary training of machine learning models offline before deployment in the vehicle. Pre-computed models are stored in memory and executed with minimal real-time computational requirements. This preliminary action separates the computationally intensive training phase from the energy-constrained inference phase, enabling accurate predictions while minimizing in-vehicle energy consumption.
Data Source
AI summary
According to embodiments, a method, for mitigating a collision between a first vehicle and a bicycle in proximity to the first vehicle, is provided. The method may include: obtaining environmental information around the first vehicle; obtaining, from the environmental information, first information associated with the bicycle; receiving, from a second vehicle, second information associated with the bicycle; predicting a trajectory of the bicycle according to the first information and the second information; determining, based on the predicted trajectory, a risk of collision between the first vehicle and the bicycle; and providing, to a driver of the first vehicle, a feedback according to the risk of collision and the predicted trajectory.