Ensuring Safe and Sustainable Autonomous Vehicle Validation on AWS with TCS Mobility Suite Smart Validation
By Vignesh Amur, Sanjay Dulepet, Apurva Vinchurkar, and Prasad Nalawade – TCS
By Karthik Thirugnanasambandam – AWS
Hundreds of thousands of lives could be saved over the next 30 years if autonomous vehicles (AVs) performed 20% better than human drivers. With capabilities like full self-driving and autopilot, autonomous vehicles can provide safe and accessible mobility for all.
On the other hand, the existing strategy of using unproven software in pilots has drawbacks. Almost 400 significant accidents involving autonomous cars occurred in the past two years, some of which resulted in fatalities. This is due to the fact that many of these cars are still under development and are conducting trials in the real world; therefore, a stronger safety case must be made before they’re used in public.
Companies are using simulation technology to solve this complex problem, but even with thorough simulation the automotive industry faces challenges with unfamiliar scenarios and edge cases to maximize the test coverage. Simulation is both an inexpensive and sustainable alternative to physical testing; however, it necessitates a large amount of compute, graphics processing units (GPUs), and power.
While cloud enables a rapid scaling of resources, it may not be an efficient use of resources to simulate everything, as there’s a need to identify the edge cases and ensure coverage.
TCS Mobility Suite Smart Validation powered by Amazon Web Services (AWS) is a solution that enables sustainable, end-to-end simulation-based validation for automated driving systems (ADS). It allows for testing of autonomous vehicles in a low-cost, risk-free manner while helping improve the quality of the software as well as accelerating time to market. Smart Validation is able to remove billions of redundant test cases, reducing them to hundreds and saving computational resources and prioritizing the testing of edge cases.
Tata Consultancy Services (TCS) is an AWS Premier Tier Services Partner and Managed Services Provider (MSP) that has been partnering with many of the world’s largest businesses in their transformation journeys for the last 50 years.
TCS Mobility Suite Smart Validation can help users design their tests, uses artificial intelligence (AI) techniques to identify edge cases and maximize test coverage, enables automated, simulator-agnostic test execution, and provides users with an analysis of their results.
Smart Validation can help expedite time to market, optimize costs, and provide flexibility by automating the testing process, reducing the need for large fleets for physical testing. It can also enable seamless switching between tools during the validation cycle. The following diagram depicts a high-level Smart Validation workflow.
Figure 1 – End-to-end workflow to define, create, execute, store, and analyze test cases.
The solution uses a variety of techniques to provide a user-friendly and effective platform for vehicle validation. Key components and features of the solution include:
- User experience (UX) and user interface (UI)-driven approach to enable a variety of stakeholders to use the tool.
- Scenario-based testing: Smart Validation provides scenarios and test cases when given an operational design domain (ODD), which focuses on what to test, and enables maximum coverage in a virtual environment X-in-the-loop, which focuses on how to test. Here, X can be software, hardware, etc.
- Behavior competency-based framework used to develop multifarious edge cases.
- Artificial intelligence: Cognitive engine uses AI and machine learning (ML) techniques to improve the identification of edge cases whilst maximizing test coverage.
- Simulation: Smart Validation middleware enables seamless, simulator-agnostic test execution.
- Test data management: Single source of truth for AV validation data.
- APIs: Smart Validation seamlessly integrates with other applications via APIs to provide users with all the relevant data in one interface.
TCS Mobility Suite Smart Validation leverages AWS services to facilitate auto-scalability and manageability of the solution. The following diagram depicts the solution’s high-level architecture.
Figure 2 – TCS Mobility Suite Smart Validation architecture.
TCS Mobility Suite Smart Validation has a web application that’s primarily used to authenticate users, configure Ego vehicles (self-driving cars that are equipped with sensors and cameras), define test suits, and behavioral competencies. A cognitive engine is used to generate test cases, while the middleware is used in conjunction with the simulator engine to execute the test cases; results are then stored in the database.
- Amazon CloudFront is used as a content delivery network (CDN) for distributing the TCS Mobility Suite Smart Validation web content with low latency and high transfer speeds across regions.
- Amazon Simple Storage Service (Amazon S3) is used to host an Angular web application with CloudFront to provide optimal UX and store test execution playback videos and logs.
- AWS App Runner is a fully managed container application service use to deploy and run the backend of the application as containers. It load balances the traffic with encryption and auto scales the underlying infrastructure without need of explicit configuration.
- Amazon DocumentDB is a fully-managed NoSQL key-value database used to store user details, vehicle, and test config information. It stores the pointers to the video files which are generated during test case execution in the simulator.
- Amazon Elastic Kubernetes Service (Amazon EKS) is a managed Kubernetes cluster service used to host the cognitive engine component. This cluster scales on the basis of the volume of test cases which needs to be generated using the cognitive engine.
- Amazon Elastic Compute Cloud (Amazon EC2) is used to host the middleware component of Smart Validation as a virtual server, taking full advantage of the flexibility EC2 offers by choosing the type of instances, operating system, CPU, and storage.
- Amazon Elastic File System (Amazon EFS) is fully managed elastic file storage used to store the solution metrics data.
- Application Load Balancer is used to enable load balancing and high availability for the various solution components hosted on EC2 instances.
- Amazon Elastic Container Service (Amazon ECS) is used to securely store, manage, and deploy application backend container images.
- Amazon CloudWatch is used for centralized logging and monitoring of the solution.
TCS Mobility Suite Smart Validation takes an innovative approach to accelerate autonomous vehicle validation, resulting in a reduction in development and validation times. This approach is more cost effective and sustainable, as it enables significant reduction in physical testing of the vehicle due to optimized simulation-based validation.
This results in improved safety through enhanced software quality, as well as increased flexibility that allows teams to switch between various simulation tools with ease throughout the development and validation cycle.
TCS Mobility Suite Smart Validation on AWS enables a more sustainable and safer approach to deploy autonomous vehicles and makes safe mobility a reality. By embracing this simulation-led and coverage-driven test framework, customers will be able to ensure thoroughness in validation testing.
TCS has a proven track record in the automotive services industry and strong collaboration with AWS to bring well-architected business solutions to market. TCS is also an AWS Automotive Competency launch partner.
TCS – AWS Partner Spotlight
TCS is an AWS Premier Tier Services Partner and MSP that has been partnering with many of the world’s largest businesses in their transformation journeys for the last 50 years.