Validation Is the Gamechanger in the SDV/AIDV Era

At VECS 2026 in Gothenburg, Dr. Manaswini Rath, Senior Vice President – SDV, Autonomous Driving & Chassis Engineering at KPIT, delivered a compelling perspective on why validation is no longer a downstream activity - it is the decisive factor in whether OEMs can deliver software-defined vehicles at speed, at quality, and at cost.

As part of the panel“Elevating Product Quality at Speed with Shift-Left Validation”, she shared real-world data, practical frameworks, and AI-powered solutions that are helping organizations rethink how they validate – from Month Zero to SOP. [linkedin.com]

The billion dollar issue: Why Validation Can't Wait

The top three North American OEMs are investing more than $7 billion to address warranty issues  – many of which trace back to defects that escaped validation and reached end users on the road. As Dr. Rath pointed out, the root cause is not a lack of infrastructure or tools. It is insufficient validation at every stage.

In SDV and multi-software integrated programs, tens of thousands of defects – 16,000, 25,000, even 80,000 – are surfacing just months before SOP. Critical-zero defects alone take 45 to 60 days to fix. When there are thousands of them, the math simply does not work within typical program timelines.

Big Bang Integration: The Real Culprit

The defect spikes are not random. It is a predictable consequence of big bang integration – where all applications, domains, and middleware are integrated at once, and validation only begins at the end.

What gets skipped:

• Platform validation – OS running on HPC/OCU (TRON OS, MBOS, Volvo OS)
• Individual domain validation – each domain tested independently
• System-level validation – interactions between HPCs, zones, and legacy systems
• Network validation – Ethernet load, wake-up behavior, battery drain

Each of these gaps contributes to the avalanche of defects at SOP.

Shift-Left = Burn Up Faster, Burn Down Earlier

Dr. Rath reframed shift-left validation as a burn-up/burn-down methodology. The goal is to accelerate defect identification early in the program (burn up) and fix them before they leak to the right side of the cycle (burn down). This prevents the late-stage defect spikes and enables faster, more predictable SOP timelines.

Three Pillars of Test Execution: Coverage, Automation & AI

The real gap is not in infrastructure or tools – it is in test execution. Three pillars define the shift:

1. Coverage – Current test coverage is often in the hundreds. What is needed is in the millions. KPIT’s QCQD framework addresses this through four dimensions:

   • Quantity – Requirements-based (7 test cases per requirement, not just 1)
   • Coverage – Standards-driven (ISO, NCAP)
   • Quality – Experience-based, human-driven boundary testing
   • Defect Reverse Engineering – Test cases derived from previously fixed defects

2. Automation – Not just scripts in pockets, but a full-stack, homogeneous validation suite  – synchronized tool chains and hardware across multi-ECU and cross-domain environments. The target: 90–99% automation.

3. AI Infusion – Multi-agent AI systems that take a Jira ticket as input and autonomously find the root cause, locate the defect in code, rewrite the code, generate test cases, run testing, and build and flash – end to end.

World Foundation Models for Autonomous Driving

For autonomous driving validation, KPIT is leveraging physical AI and deep learning-based world foundation models to generate millions of synthetic driving scenes – indistinguishable from real footage and aligned to NCAP and ISO standards. As Dr. Rath put it:“The whole game is how do you make the simulation a reality.”

2-Year SOP Is Already Happening

When asked about balancing aggressive 2-year SOP timelines with safety demands, Dr. Rath confirmed this is not theoretical – KPIT is already doing it with OEMs. Two to three vehicle variants are launching in Q1 2027 on a 2-year SOP plan. The key: validation starts from Month Zero, with ready solutions – pre-built test case libraries, standardized formats (XODR), and pre-configured infrastructure.

Architecture Must Be Designed for Testability

SDV architecture must be validated during the architecture phase itself. KPIT’s group company Technica Engineering provides a unique E/E architecture validation bench that emulates network performance and whole vehicle performance with dummy loads – catching architecture-level issues before they become SOP killers.

The Bottom Line

Speed + Quality + Cost = Coverage + Automation + AI

The answer to achieving “China speed” SOP is not just better infrastructure – it is fundamentally about improving test execution through massive coverage, near-complete automation, and AI-powered acceleration, starting from Day Zero.

Watch the full session:

Learn more about KPIT’s Validation expertise: kpit.com/expertise/validation

Author:

Dr. Manaswini Rath
Senior Vice President, Global Strategy & Solutioning, KPIT