Bicycle Intention Detection Using Visual Cues and Sensor Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing autonomous vehicle systems struggle to accurately predict the future intentions of bicycles and similar person-wide vehicles, particularly due to the difficulty in detecting visual cues from LIDAR data alone.
Innovation Solution
The use of visual data from cameras and image sensors to input into a machine-learned model, which predicts the future intentions of bicycles by identifying visual cues such as head turning, operation of controls, and leaning, in addition to LIDAR, radar, or other sensor data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If LIDAR data alone is used to detect bicycles, then the system complexity is reduced, but the measurement precision of future intention detection deteriorates
Solution Approach 1:
The patent combines LIDAR data with visual data from cameras to detect bicycles and predict their future intentions. The LIDAR provides accurate spatial positioning and depth information, while the camera captures visual cues like head turning, body leaning, and control operations. This merging of multiple data sources resolves the contradiction by maintaining system complexity while significantly improving intention detection precision through multi-modal fusion.
Solution Approach 2:
The patent introduces a machine-learned model as an intermediary that processes and fuses LIDAR data with visual data from cameras. This intermediary component analyzes visual cues such as head orientation, body lean angle, and control handle operations to predict future bicycle intentions. The machine-learned model acts as a mediator that transforms raw multi-source data into actionable intention predictions, resolving the precision-complexity contradiction.
2Measurement precision
If visual data from cameras is added to LIDAR data, then the measurement precision of intention detection is improved, but the device complexity increases
Solution Approach 1:
The patent implements a machine-learned model that serves multiple functions: it processes LIDAR data, analyzes visual data from cameras, detects various visual cues (head turning, body leaning, control operations), and predicts future bicycle intentions. This multi-functional approach improves intention detection precision while managing system complexity by consolidating multiple processing tasks into a single intelligent system rather than requiring separate dedicated systems for each function.
3Reliability
If machine-learned models are used to predict future intentions, then the reliability of autonomous vehicle navigation is improved, but the loss of processing time increases
Solution Approach 1:
The patent uses the machine-learned model to predict future bicycle intentions in advance of actual maneuvers by analyzing current visual cues such as head turning direction, body lean angle, and control handle position. This preliminary prediction allows the autonomous vehicle to prepare navigation decisions before the bicycle actually executes its maneuver, improving navigation reliability while minimizing processing time delays by performing predictions proactively rather than reactively.
Data Source
AI summary
There is provided methods, systems, and computer-readable media for determining intention of bicycles and other person-wide vehicles. A method comprises receiving, from a first sensor of an autonomous vehicle, first sensor data relating to an external environment of the autonomous vehicle; and receiving, from a second sensor of the autonomous vehicle, second sensor data the second sensor comprising a different sensor type to the first sensor. A person-wide vehicle in proximate to the autonomous vehicle is identified. First object data associated with the person-wide vehicle is determined. Based on the first object data, a future intention of the person-wide vehicle is received from a machine-learned model. The autonomous vehicle is controlled based at least in part on the future intention of the person-wide vehicle.