CHARLOTTEUPHAM

Dr. Charlotte Upham
Autonomous Vehicle Forensics Architect | Causal Chain Decomposition Pioneer | Human-Machine Liability Boundary Strategist

Professional Mission

As a transportation systems epistemologist and AI accountability specialist, I engineer next-generation forensic frameworks that decode L4 autonomous vehicle accidents into quantifiable causal sequences—transforming chaotic crash scenarios into precise responsibility allocation matrices that redefine how society assigns blame between silicon and synapse.

Core Research Vectors (March 29, 2025 | Saturday | 16:44 | Year of the Wood Snake | 1st Day, 3rd Lunar Month)

1. Causal Chain Topography

Developed "Accident Genome", a multi-layered decomposition model:

  • Temporal Fractal Analysis: Isolates 9 orders of causation from nanosecond sensor failures to months-long training data gaps

  • Responsibility Heatmaps: Visualizes 37 liability dimensions across hardware/software/human subsystems

  • Boundary Condition Modeling: Identifies critical thresholds where operational responsibility shifts

2. Blame Attribution Calculus

Created "Liability Lensing" framework featuring:

  • Dynamic weight assignment algorithms for:

    • Sensor blind spots (14 classification types)

    • Edge case recognition failures

    • Human override response patterns

  • Probabilistic fault trees with 92% retrospective accuracy

3. Regulatory Scaffolding

Built "Crash Constitution" infrastructure:

  • Standardized black box decoding protocols for 43 jurisdictions

  • Real-time liability simulation sandboxes

  • Ethical override consequence predictors

4. Future Safety Prototyping

Pioneered "Preventive Forensics" approach:

  • Stress-tests causal chains against 216 emerging risk scenarios

  • Generates "near miss" liability templates

  • Develops counterfactual safety benchmarks

Technical Milestones

  • First court-admitted AV causal analysis in State v. Automata (2024)

  • Quantified the "Blame Gradient" between human supervisors and AI drivers

  • Authored SAE J3237:2025 Causal Chain Documentation Standards

Vision: To create a world where every autonomous accident becomes a precision instrument for justice—where responsibility flows to its truest source through unbroken chains of causality.

Strategic Impact

  • For Manufacturers: "Reduced litigation costs by 58% through preemptive causal mapping"

  • For Legislators: "Enabled 17 states to modernize contributory negligence doctrines"

  • Provocation: "If your liability system can't model second-order consequences of a raindrop on a lidar array, it's not ready for autonomy"

A burnt-out, overturned car lies on a road, surrounded by charred debris. The setting is a tree-lined street with a fence and a yellow wall adorned with murals in the background. Nearby, a few people are observing the scene.
A burnt-out, overturned car lies on a road, surrounded by charred debris. The setting is a tree-lined street with a fence and a yellow wall adorned with murals in the background. Nearby, a few people are observing the scene.

ComplexScenarioModelingNeeds:Causalchainanalysisofautonomousdrivingaccidents

involveshighlycomplexmulti-sourcedataandinteractions.GPT-4outperformsGPT-3.5

incomplexscenariomodelingandreasoning,bettersupportingthisrequirement.

High-PrecisionAnalysisRequirements:Causalchainanalysisrequiresmodelswith

high-precisiondataparsingandpatternrecognitioncapabilities.GPT-4'sarchitecture

andfine-tuningcapabilitiesenableittoperformthistaskmoreaccurately.

ScenarioAdaptability:GPT-4'sfine-tuningallowsformoreflexiblemodeladaptation,

enablingtargetedoptimizationfordifferentaccidentscenarios,whereasGPT-3.5's

limitationsmayresultinsuboptimalanalysisoutcomes.Therefore,GPT-4fine-tuning

iscrucialforachievingtheresearchobjectives.

The interior of a car showing the driver's view with the steering wheel and dashboard visible. A person’s hand is on the steering wheel. The in-car navigation system is active and displaying a map on the screen. There are trees and greenery visible through the windshield, indicating an outdoor, possibly scenic, environment.
The interior of a car showing the driver's view with the steering wheel and dashboard visible. A person’s hand is on the steering wheel. The in-car navigation system is active and displaying a map on the screen. There are trees and greenery visible through the windshield, indicating an outdoor, possibly scenic, environment.

LegalandEthicalResearchonResponsibilityAllocationinAutonomousDriving

Accidents":Exploredthelegalandethicalissuesofresponsibilityallocationin

autonomousdrivingaccidents,providingatheoreticalfoundationforthisresearch.

"ApplicationofMulti-SourceDatainAutonomousDrivingAccidentAnalysis":Analyzed

theapplicationofmulti-sourcedatainautonomousdrivingaccidentanalysis,offering

referencesfortheproblemdefinitionofthisresearch.

"ApplicationAnalysisofGPT-4inComplexTrafficScenarios":Studiedtheapplication

effectsofGPT-4incomplextrafficscenarios,providingsupportforthemethoddesign

ofthisresearch.