AI-Powered Transportation Planning and Safety Countermeasure Visualization

inrix · Government · 2024 · source Agentic: no · promotional bias: med · source-backed · extracted with prompt 0.5.4 (schema v2)

Generated from extraction.json — do not edit by hand. Every claim cites verbatim source evidence; stated means the quote directly asserts the claim, implied means it supports it indirectly.

Problem

Domain: Transportation planning and road-safety countermeasures

Transportation agencies need to identify high-risk locations for vulnerable road users and produce safety countermeasure concepts faster than traditional manual planning and visualization workflows allow.

Constraints:

transportation data spans connected cars, mobile devices, roadway sensors, events, crashes, assets, and points of interest
recommendations must be grounded in transportation safety knowledge and local conditions
conceptual drawings normally require multiple specialist teams and review cycles
visual outputs need to incorporate city-specific rules, local standards, and community considerations
the public case is a proof-of-concept/vendor blog and does not provide production adoption metrics

Architecture

INRIX Compass combines INRIX's transportation data lake with Amazon Bedrock, Bedrock Knowledge Bases, Anthropic Claude, API Gateway, EKS, Lambda, and Amazon Nova Canvas. The countermeasure pipeline answers natural-language questions over roadway risk using RAG and foundation models, generating prioritized safety recommendations for selected locations. The visualization pipeline takes those countermeasures and uses Nova Canvas text-to-image, image-to-image, and in-painting capabilities to produce conceptual street-view drawings for review.

Diagram confidence: explicit (topology described by the source).

flowchart LR
  inrix-compass["inrix-compass"] --> bedrock-rag-layer["bedrock-rag-layer"]
  inrix-compass["inrix-compass"] --> countermeasure-generation-service["countermeasure-generation-service"]
  inrix-transportation-data-lake["inrix-transportation-data-lake"] --> bedrock-rag-layer["bedrock-rag-layer"]
  inrix-transportation-data-lake["inrix-transportation-data-lake"] --> countermeasure-generation-service["countermeasure-generation-service"]
  bedrock-rag-layer["bedrock-rag-layer"] --> countermeasure-generation-service["countermeasure-generation-service"]
  countermeasure-generation-service["countermeasure-generation-service"] --> api-gateway-eks-layer["api-gateway-eks-layer"]
  countermeasure-generation-service["countermeasure-generation-service"] --> nova-canvas-visualizer["nova-canvas-visualizer"]
  api-gateway-lambda-layer["api-gateway-lambda-layer"] --> countermeasure-generation-service["countermeasure-generation-service"]
  api-gateway-lambda-layer["api-gateway-lambda-layer"] --> nova-canvas-visualizer["nova-canvas-visualizer"]
  nova-canvas-visualizer["nova-canvas-visualizer"] --> review-stakeholders["review-stakeholders"]

Drag to pan · scroll or use +/− to zoom · reconstructed from the source's described component topology

Components

Component	Role	Interacts with
`inrix-compass`	User-facing transportation intelligence application for natural-language risk and countermeasure questions.	`bedrock-rag-layer`, `countermeasure-generation-service`
`inrix-transportation-data-lake`	Large real-time and historical transportation data store used for risk and roadway analysis.	`bedrock-rag-layer`, `countermeasure-generation-service`
`bedrock-rag-layer`	Amazon Bedrock Knowledge Bases and Claude models implementing retrieval-augmented generation over roadway data and safety knowledge.	`inrix-compass`, `inrix-transportation-data-lake`, `countermeasure-generation-service`
`countermeasure-generation-service`	Generates prioritized operational and design safety countermeasures for selected locations.	`bedrock-rag-layer`, `api-gateway-eks-layer`, `nova-canvas-visualizer`
`api-gateway-eks-layer`	Manages API requests and responses for the countermeasure generation path.	`countermeasure-generation-service`
`api-gateway-lambda-layer`	Processes visualization requests between API Gateway, Lambda, and Bedrock.	`countermeasure-generation-service`, `nova-canvas-visualizer`
`nova-canvas-visualizer`	Generates and edits conceptual street-view countermeasure images using Amazon Nova Canvas.	`api-gateway-lambda-layer`, `review-stakeholders`
`review-stakeholders`	Transportation engineers, planners, landscape architects, safety analysts, public works, and traffic operations reviewers.	`nova-canvas-visualizer`

Data flow

Users ask Compass for risky locations or countermeasures. INRIX Compass queries the roadway network and data lake through Bedrock Knowledge Bases and Claude-backed RAG, identifies systemic risk factors and anomalous safety patterns, and returns prioritized operational or design countermeasures. Those countermeasures flow through API Gateway/EKS and into the visualization path, where API Gateway/Lambda process Bedrock requests and Nova Canvas generates or edits street-view concepts.

4 supporting quotes

“INRIX Compass generates the countermeasures for a selected location * Amazon API Gateway and Amazon Elastic Kubernetes Service (Amazon EKS) manage API requests and responses”article, stated

“/) (Amazon EKS) manage API requests and responses * Amazon Bedrock Knowledge Bases and Anthropic’s Claude Models provide [Retrieval Augmented Generation (RAG)](https://aws.amazon.com/what-is/retrieval-”article, stated

“Behind the scenes, Compass AI uses RAG and Amazon Bedrock powered foundation models (FMs) to query the roadway network to identify and prioritize locations with systemic risk factors and anomalous safety patterns.”article, stated

“Amazon Bedrock with model access to Amazon Nova Canvas provide image generation and in-painting”article, stated

Why this architecture

INRIX split the workflow into recommendation and visualization stages because transportation planning requires both evidence-based safety countermeasures and reviewable conceptual drawings. RAG over Compass and Bedrock grounds risk analysis in roadway data and safety knowledge, while Nova Canvas accelerates the specialized visualization loop. API Gateway, EKS, and Lambda provide managed request handling around the model calls.

3 supporting quotes

“This solution uses INRIX Compass countermeasures as the input, AWS serverless architecture, and Amazon Nova Canvas as the image visualizer.”article, stated

“The generation of countermeasure suggestions represents the initial phase in transportation planning.”article, stated

“Image visualization requires the crucial next step of preparing conceptual drawings.”article, stated

Models used: Amazon Bedrock · Amazon Bedrock Knowledge Bases · Anthropic Claude · Amazon Nova Canvas · Amazon API Gateway · Amazon EKS · AWS Lambda

Scale

INRIX has used GPS data from connected vehicles for more than 20 years.
“For over 20 years, INRIX has been a leader for probe-based connected vehicle and device data and insights”article, stated
The proof of concept was selected by Caltrans in June 2024.
“In June 2024, the State of California’s Department of Transportation (Caltrans) selected INRIX for a proof of concept for a generative AI-powered solution to improve safety for vulnerable road users (VRUs).”article, stated
INRIX Compass uses INRIX's 50 PB data lake.
“INRIX Compass is an application that harnesses generative AI and INRIX’s 50 PB data lake to solve transportation challenges.”article, stated
The example natural-language query asks for the top five highest-risk locations.
“Where are the top five locations with the highest risk for vulnerable road users?”article, stated
The visualization workflow is claimed to reduce conceptual design cycles from weeks to days.
“potentially reducing the design cycle from weeks to days.”article, stated

Alternatives considered (roads not taken)

Traditional manual conceptual drawing workflow

Rejected because: It requires many specialized teams, multiple review cycles, and delays safety countermeasure deployment.

“This process has traditionally been time-consuming due to the involvement of multiple specialized teams”article, stated

“Consequently, this lengthy and costly process frequently leads to delays in implementing safety countermeasures.”article, stated

Countermeasure text without visual concepts

Rejected because: Transportation planning also needs conceptual drawings that stakeholders can review for feasibility, design, safety, and local compliance.

“Image visualization requires the crucial next step of preparing conceptual drawings.”article, stated

“local regulations, environmental considerations, and community feedback must be incorporated into the design process.”article, stated

Generic image generation without local references

Rejected because: The prototype uses few-shot reference images and prompts so generated concepts can reflect city-specific requirements.

“the system incorporates a few-shot learning approach with reference images and carefully crafted prompts”article, stated

“allowing for seamless integration of city-specific requirements into the generated outputs.”article, stated

Patterns

`rag-retrieval` (retrieval)

Compass uses Bedrock Knowledge Bases and Claude models to ground natural-language roadway-risk questions in transportation data and safety knowledge.

`tool-use-orchestration` (workflow)

The workflow orchestrates risk query, countermeasure generation, API request handling, and image generation services rather than asking one model to do everything.

“Key components include: * Countermeasures generation: * INRIX Compass generates the countermeasures for a selected location”article, stated

“0ARIsAFb5Ac9JNeng4ci9WDMGQyes6pqvmBtUadHExWZMWO0Gmlh0bOOfXzAWUD0aArU6EALw_wcB) process requests from API Gateway and Amazon Bedrock * Amazon Bedrock with model access to Amazon Nova Canvas provide image generation and in-painting”article, stated

`synthetic-data-generation` (generation)

Amazon Nova Canvas generates and edits countermeasure visualizations from text and reference images.

“Amazon Nova Canvas delivers advanced image processing through text-to-image generation and image-to-image transformation capabilities.”article, stated

`structured-output-contract` (reliability)

The system separates countermeasure generation from conceptual-image rendering, which constrains the model's role in each stage.

“This solution uses INRIX Compass countermeasures as the input, AWS serverless architecture, and Amazon Nova Canvas as the image visualizer.”article, stated

Tradeoffs

Use generative image tooling for concept acceleration.

Chose Nova Canvas prototyping and in-painting over Only traditional CAD/concept drawing cycles.

Rationale: Generated concepts can be iterated and reviewed faster, potentially reducing design cycles from weeks to days.

“This prototyped solution enables rapid iteration of conceptual drawings that can be efficiently reviewed by various teams”article, stated

Ground recommendations in RAG over transportation data and manuals.

Chose Compass plus Bedrock Knowledge Bases over Ungrounded foundation-model recommendations.

Rationale: Safety countermeasures need roadway conditions, crash/asset/POI data, industry knowledge, and domain manuals.

“Trained on real-time and historical data and industry research and manuals, the solution provides a new systemic, safety-based methodology”article, stated

Treat this as a proof of concept.

Chose Prototype for Caltrans-selected safety use case over Claiming broad production deployment.

Rationale: The article describes a selected proof of concept and prototype, not a fully deployed agency-wide system.

“In June 2024, the State of California’s Department of Transportation (Caltrans) selected INRIX for a proof of concept for a generative AI-powered solution to improve safety for vulnerable road users (VRUs).”article, stated

Gotchas (what bit them)

Transportation planning has too many stakeholder checks for a single-shot text answer.

Root cause: Engineering feasibility, urban planning, safety analysis, public works, traffic operations, environmental rules, and community feedback all affect countermeasure design.

Resolution: Use AI to accelerate conceptual drawings but keep outputs reviewable by specialist teams.

“These teams work collaboratively, creating and iteratively refining various visualizations based on feedback from urban designers and other stakeholders.”article, stated

“Each iteration cycle typically involves multiple rounds of reviews, adjustments, and approvals”article, stated

Generated visuals need local-context controls.

Root cause: City-specific rules, design requirements, local regulations, environmental considerations, and community feedback differ by jurisdiction.

Resolution: Use few-shot reference images and carefully crafted prompts to incorporate city-specific requirements.

“The complexity is further amplified by city-specific rules and design requirements, which often necessitate significant customization.”article, stated

“the system incorporates a few-shot learning approach with reference images and carefully crafted prompts”article, stated

Evaluation

The case evaluates feasibility through a Caltrans proof of concept, safety-risk prioritization, generated countermeasure recommendations, reviewable conceptual visuals, and claimed reduction of visualization cycle time from weeks to days. It does not publish crash-reduction or production-adoption results.

Metrics: high-risk location identification · countermeasure recommendation quality · conceptual drawing cycle time · local-standard compliance · proof-of-concept feasibility

3 supporting quotes

“risk assessment, location prioritization, and project implementation.”article, stated

“This approach not only accelerates the design process but also supports consistency across different projects while maintaining compliance with local standards.”article, stated

Outcomes

INRIX and AWS demonstrated a Bedrock-backed workflow for vulnerable-road-user safety analysis.
The system can answer natural-language questions about high-risk locations and countermeasures.
Nova Canvas can generate conceptual countermeasure visualizations for stakeholder review.
The writeup claims visualization cycles can potentially move from weeks to days.
The case does not report deployed crash-safety outcomes.

The proof of concept uses a 50 PB INRIX data lake and example top-five risk-location queries, but production usage volume is not disclosed.