Event-Based Object Detection for Autonomous Vehicle Time-to-Contact
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous driving systems face challenges in accurately detecting and responding to adversarial objects, particularly slow or static objects that can lead to collisions due to limitations in existing object detection and time-to-contact calculation methods.
Innovation Solution
The implementation of a Dynamic Vision Sensor (DVS) and a convolutional neural network (CNN) for fast and accurate time-to-contact calculation, combined with an event-based camera system that uses a configurable light source to enhance detection of adversarial objects, particularly vulnerable road users, by filtering events based on a specific frequency range.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing object detection methods are used, then the system can detect common objects, but it fails to accurately detect slow or static adversarial objects leading to collisions
Solution Approach 1:
The system performs preliminary actions by projecting a light pattern onto the environment before detecting objects. This active illumination allows the sensor to detect objects that would otherwise be invisible or indistinguishable from the background, particularly slow or static adversarial objects that traditional passive detection methods miss.
Solution Approach 2:
The patent introduces an intermediary light projection system between the sensor and objects. The light pattern acts as a mediator that enhances object visibility by creating contrast patterns on detected objects, allowing the sensor to distinguish adversarial objects from the background more reliably.
2Loss of time
If traditional sensors are used, then the system can operate without additional components, but it cannot provide timely detection of adversarial objects
Solution Approach 1:
The patent merges the light projection function and object detection function into a single integrated system. The same device that captures images also projects light patterns, eliminating the need for separate illumination hardware and reducing system complexity while enabling timely detection of adversarial objects.
Solution Approach 2:
The sensor device is designed to perform multiple functions: it acts as both an image capturer and a light projector. This multi-functionality allows the system to detect adversarial objects in real-time without adding separate dedicated components, reducing overall system complexity while improving detection speed.
3Reliability
If the system uses active light projection, then it can detect adversarial objects, but it increases energy consumption
Solution Approach 1:
The light projection operates periodically rather than continuously, projecting light patterns at specific intervals to detect objects. This periodic operation reduces energy consumption compared to continuous illumination while still maintaining reliable detection of adversarial objects by capturing images at strategically timed moments when light patterns are projected.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enables robust and efficient detection of adversarial objects, improving the safety and reliability of autonomous vehicles by providing timely and accurate time-to-contact information, reducing the risk of collisions with unexpected obstacles.
Implementation Method 1
obtaining a plurality of event data associated with a change in a light intensity of at least one pixel
Implementation Method 2
a convolutional neural network (CNN) for fast and accurate time-to-contact calculation
Implementation Method 3
an event-based camera system that uses a configurable light source to enhance detection of adversarial objects, particularly vulnerable road users, by filtering events based on a specific frequency range
Data Source
AI summary
A method for calculating a time to contact of an autonomous vehicle, the method comprising: obtaining a plurality of event data an image, wherein the event data is associated with a pixel associated with a change in light intensity; determining a reference signal frequency associated with a transmitted light; identifying a select event data from the plurality of event data, wherein the light frequency associated with the select event data is substantially the same as the reference signal frequency; determining an object based on the select event data, wherein the object is fully enclosed by a bounding box comprising coordinates of a rectangular border; calculating a distance between a set of coordinates of the bounding box closest to the autonomous vehicle and the autonomous vehicle; and calculating the time to contact between the set of coordinates of the bounding box and the autonomous vehicle.