Spiking Reichardt Detector for Low-Power Collision Detection

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing collision detection methods for small robots, such as micro UAVs, are inefficient due to high power consumption and inability to distinguish between approaching, receding, or translating objects using a single lightweight front camera, especially in complex environments.

Innovation Solution

A low-power spiking neural network implementation of the Reichardt motion detector that divides images into visual sub-fields, using a spiking Reichardt detector to detect motion and inhibit signals, accumulating spikes to determine impending collisions, and distinguishing between approaching, receding, or translating objects.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If a traditional Reichardt motion detector is used to detect motion in all directions and sum them, then motion detection coverage is improved, but the ability to distinguish between approaching, receding, or translating objects deteriorates

Engineering Contradiction:

Improvemotion detection coverageVSAvoidobject motion distinction accuracy

Core Design Contradiction:

Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The visual field is segmented into multiple directional sub-fields (e.g., left, right, up, down), with each sub-field dedicated to detecting motion in a specific direction. This segmentation allows the system to maintain comprehensive motion detection coverage while precisely determining the direction of motion, thereby resolving the contradiction between coverage and precision.

Inventive Principle:

Principle #1Segmentation

2Reliability

If multiple cameras are used to improve collision detection accuracy, then detection reliability is improved, but device weight and power consumption worsen

Engineering Contradiction:

Improvecollision detection accuracyVSAvoidrobot weight

Core Design Contradiction:

ReliabilityVSWeight of moving object

Solution Approach 1:

Instead of using multiple cameras, the system segments the visual field of a single camera into multiple directional sub-fields. Each sub-field processes motion information for a specific direction, enabling accurate collision detection without the need for additional cameras, thus maintaining reliability while reducing weight.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

A single camera is made multi-functional by processing different directional information through various visual sub-fields simultaneously. This allows one camera to perform the function that would otherwise require multiple cameras, reducing overall system weight while maintaining detection accuracy.

Inventive Principle:

Principle #6Universality (Multi-functionality)

3Weight of moving object

If a single camera is used to reduce weight, then device weight is improved, but the ability to distinguish object motion directions deteriorates

Engineering Contradiction:

Improverobot weightVSAvoidmotion direction distinction accuracy

Core Design Contradiction:

Weight of moving objectVSMeasurement precision

Solution Approach 1:

The single camera's visual field is divided into multiple directional sub-fields, each specialized for detecting motion in a particular direction. This segmentation enables the single camera to accurately distinguish motion directions without requiring multiple cameras, thus maintaining lightweight design while improving measurement precision.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system adds a directional dimension to the single camera's output by creating multiple visual sub-fields with different directional sensitivities. This dimensional transformation allows the single camera to provide multi-directional motion information that would otherwise require multiple cameras.

Inventive Principle:

Principle #17Another dimension (Dimensionality change)

4Reliability

If complex processing algorithms are used to improve collision detection accuracy, then detection reliability is improved, but power consumption worsens

Engineering Contradiction:

Improvecollision detection accuracyVSAvoidpower consumption

Core Design Contradiction:

ReliabilityVSUse of energy by moving object

Solution Approach 1:

The processing algorithm is segmented into simple directional motion detectors, each handling a specific direction. This segmentation replaces complex algorithms with multiple simple, specialized detectors, reducing overall computational complexity and power consumption while maintaining detection accuracy.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system changes the processing parameter from complex general-purpose motion analysis to simple directional motion detection in each sub-field. This parameter change simplifies the computational requirements while maintaining the ability to accurately detect collision-related motion.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS9934437B1System and method for real-time collision detection

Publication Date: 2018.04.03 HRL LAB

AI summary

Described is a system for collision detection. The system divides an image in a sequence of images into multiple sub-fields comprising complementary visual sub-fields. For each visual sub-field, motion is detected in a direction corresponding to the visual sub-field using a spiking Reichardt detector with a spiking neural network. Motion in a direction complementary to the visual sub-field is also detected using the spiking Reichardt detector. Outputs of the spiking Reichardt detector, comprising data corresponding to one direction of movement from two complementary visual sub-fields, are processed using a movement detector. Based on the output of the movement detector, an impending collision is signaled.