Trajectory Reference Frame Alignment for Sensor Accuracy Validation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing approaches for collecting prior trajectories of agents using lower-fidelity sensor systems face challenges in aligning different temporal and spatial reference frames, making it difficult to evaluate and validate the accuracy of trajectory representations derived from these systems.

Innovation Solution

A technique for aligning different representations of an agent's trajectory by applying spatial and temporal transformations to reconcile differences in reference frames, allowing for accurate comparison and validation of lower-fidelity sensor data against higher-fidelity data without physical modifications or calibrations.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If lower-fidelity sensor systems are used to collect trajectory data, then cost and scalability are improved, but alignment accuracy with higher-fidelity reference frames deteriorates

Engineering Contradiction:

Improvecost-effectivenessVSAvoidtrajectory alignment accuracy

Core Design Contradiction:

Ease of manufactureVSMeasurement precision

Solution Approach 1:

The patent introduces intermediary transformation models (spatial transformation model and temporal transformation model) that act as mediators between lower-fidelity sensor data and higher-fidelity reference frames. These models enable accurate alignment without requiring physical calibration hardware or complex modifications to the sensor systems themselves, thus maintaining cost-effectiveness while improving measurement precision.

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent transforms the alignment problem from a physical calibration task to a parameter optimization task. By adjusting transformation parameters (spatial transformation parameters and temporal transformation parameters) through optimization algorithms, the system achieves accurate alignment between different fidelity sensor systems without requiring physical modifications, resolving the contradiction between cost and precision.

Inventive Principle:

Principle #35Parameter changes

2Measurement precision

If physical modifications or calibrations are applied to align sensor reference frames, then alignment accuracy is improved, but system complexity and deployment difficulty increase

Engineering Contradiction:

Improvereference frame alignment accuracyVSAvoidcalibration system complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent replaces mechanical/physical calibration systems with computational transformation models. Instead of using physical modifications, calibration hardware, or complex mechanical alignment procedures, the system uses software-based spatial and temporal transformation models that can be applied algorithmically, significantly reducing system complexity while maintaining alignment accuracy.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent creates virtual copies of reference frames through transformation models rather than requiring physical calibration artifacts. The spatial transformation model generates transformed representations of reference frames that can be digitally aligned with sensor data, eliminating the need for physical calibration objects and reducing deployment complexity.

Inventive Principle:

Principle #26Copying

3Adaptability or versatility

If different sensor systems with different reference frames are used, then data collection versatility is improved, but data integration and comparison difficulty increase

Engineering Contradiction:

Improvesensor system versatilityVSAvoiddata integration complexity

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent creates a universal transformation framework that can handle multiple types of sensor systems with different reference frames. The spatial transformation model and temporal transformation model are designed to work with various sensor modalities (lidar, cameras, radar) and reference frame configurations, enabling versatile data collection while simplifying integration through a unified transformation approach.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS12555252B2Systems and methods for aligning trajectory information derived from different source data

Publication Date: 2026.02.17 LYFT INC

AI summary

Examples disclosed herein may involve a computing system that is operable to (i) use a first approach to produce a first representation of an agent's trajectory from a first set of sensor data, (ii) use a second approach to produce a second representation of the agent's trajectory from a second set of sensor data, wherein the first and second representations of the agent's trajectory are based on different spatial reference frames and different temporal reference frames, (iii) align the spatial reference frames of the first and second representations by applying a spatial transformation to one of the first or second representations, (iv) align the temporal reference frames by applying an origin-time offset to one of the first or second representations, and (v) use the aligned first and second representations as a basis for evaluating an accuracy of the first approach relative to the second approach.