Optical Flow Cancellation via Spiking Neural Network Pulse Latency Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing optical flow detection methods in robotics require substantial computational resources and multiple sensors, leading to noisy results and the inability to distinguish between close, slowly moving objects and distant, faster-moving objects effectively.
Innovation Solution
A computer-implemented method using an artificial spiking neuron network to encode optical flow into pulse latency, preventing pulse generation based solely on robotic motion and eliminating self-motion-induced optical flow components, allowing for accurate object detection and distance estimation with a single camera.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If differential methods are used for optical flow estimation, then object detection capability is improved, but computational resource requirements increase substantially
Solution Approach 1:
The patent extracts and eliminates the self-motion component from the optical flow field, separating it from the object motion component. This is achieved by encoding self-motion information into a cancellation signal that removes the platform-induced flow patterns, leaving only the object motion information that needs to be processed
Solution Approach 2:
The patent introduces an intermediary self-motion encoder that processes platform motion information and generates a cancellation signal. This intermediary component mediates between the raw optical flow and the object detection process, filtering out the harmful self-motion component before it reaches the differential method processor
2Measurement precision
If multiple sensors are used to distinguish between close slowly moving objects and distant faster-moving objects, then object classification accuracy is improved, but device complexity increases
Solution Approach 1:
The patent transforms the optical flow information into a different dimensional representation by encoding it into pulse latencies in a spiking neural network. This temporal encoding dimension allows the system to distinguish object properties without requiring additional spatial sensors, as the latency patterns encode both distance and velocity information
3Productivity
If optical flow cancellation is implemented using spiking neural networks, then computational requirements are reduced, but measurement precision may be affected
Solution Approach 1:
The patent replaces the traditional mechanical/differential computation system with a spiking neural network system that uses temporal pulse patterns. This substitution maintains measurement precision while dramatically reducing computational requirements, as the spiking network performs optical flow estimation through natural temporal coding rather than intensive algebraic computation
Data Source
AI summary
Optical flow for a moving platform may be encoded into pulse output. Optical flow contribution induced due to the platform self-motion may be cancelled. The cancellation may be effectuated by (i) encoding the platform motion into pulse output; and (ii) inhibiting pulse generation by neurons configured to encode optical flow component optical flow that occur based on self-motion. The motion encoded may be coupled to the optical flow encoder via one or more connections. Connection propagation delay may be configured during encoder calibration in the absence of obstacles so as to provide system specific delay matrix. The inhibition may be based on a coincident arrival of the motion spiking signal via the calibrated connections to the optical flow encoder neurons. The coincident motion pulse arrival may be utilized in order to implement an addition of two or more vector properties.