Partnership & Integration

Shadow Mode POC —
Zero Interference

ADD runs in parallel with your existing stack — passive, non-interfering. It has been validated across 2,262 simulation scenarios. The next step is real-world data — and we want to do that with you.

POC Process

How an Evaluation Works

1

Introductory Email Exchange

We start with a brief email exchange to understand your current ADS stack, perception output fields, and evaluation goals. Based on your perception schema, we adapt the bridge layer to your data format — you don't rewrite your stack to fit ours.

Async · Email
2

Shadow Mode Deployment

ADD is deployed alongside your existing system. It receives the same perception data and produces its own decision output — no connection to vehicle actuators. Your current system remains in sole control.

1–3 days · On-site or remote
3

Scenario Evaluation

Run ADD Studio against your scenario library — or build new scenarios on the spot using the Scene Editor. The SDS Test Runner outputs side-by-side decision comparison, latency benchmarks, and full traceability logs for each case.

For teams running ADD on real-world perception data for the first time, this evaluation becomes a co-validation — your data strengthens the system, and your team gets first-mover insight into its real-world behaviour.

1–2 weeks · Your data
4

Results & Next Steps

Structured report covering decision accuracy, latency benchmarks, jurisdiction switching results, and ISO 26262 traceability output. Presented to your technical leadership — no commitment required.

Joint review · No obligation

Typical scoping questions we ask

  • What technology route does your ADS planning layer currently use? (Rule-based / E2E / hybrid)
  • What fields does your perception layer currently output? What is the bridging effort to align with ADD?
  • Which scenario categories are highest priority? (Regional traffic law / Vulnerable road users / Multi-vehicle conflict)
  • What is your target version for the first delivery phase — Basic, Pro, or Plus?
Who This Is For

Integration Profiles

OEM System Teams

Replacing or augmenting an existing decision layer with a certifiable, white-box alternative ahead of a production programme milestone or 2026 safety regulation deadline.

Tier-1 Suppliers

Adding ADD as the pre-constraint layer to an existing ADAS or AD platform — improving ISO 26262 readiness and reducing type-approval documentation overhead.

FAQ

Common Questions

Does shadow mode affect our existing system? +
No. Shadow mode is fully passive. ADD receives the same perception data as your system and produces parallel decision outputs, but has no connection to vehicle actuators. Your existing system remains in sole control throughout the evaluation.
What sensor stack or perception format do we need? +
ADD works with your existing perception stack — we adapt to your format, not the other way around. Share your perception output schema with us — field names, data types, coordinate conventions — and we align the bridge layer to match. You don't rewrite your stack to fit ours. The interface specification document we provide during scoping describes exactly what ADD needs, and we do the mapping work on our side.
How long does a typical POC take? +
The initial email exchange typically takes 2–3 days. From there, shadow mode deployment is usually complete within 1–3 working days, and the full evaluation to final report runs 2–4 weeks in total.
Is source code shared during a POC? +
No. ADD is deployed as a compiled runtime module. The rule library is provided in human-readable YAML form so your safety engineers can review decision logic, but the underlying engine implementation remains proprietary. A full NDA is executed before any technical materials are shared.
Can ADD work alongside our E2E neural planning layer? +
Yes. ADD replaces only the Safety Check layer in a conventional E2E pipeline. The planning layer retains full implicit generalisation capability; ADD adds a white-box, auditable constraint space around it. OEM teams regain control over driving policy design without returning to the supplier for model retraining.
What does 2026 safety regulation support look like? +
ADD's responsibility chain traceability is verified at 0.18ms and outputs structured audit logs for direct ingestion by safety validation tools. We provide a Safety Case Summary document and can support your ASIL decomposition and SOTIF analysis. Full documentation available under NDA to qualified organisations.
Has ADD been tested on real vehicle data? +
ADD is currently simulation-validated across 2,262 test items and 10 core driving scenarios — all passing. Real-world perception data integration is the explicit goal of our early partner programme. If you bring your stack, we bring the engine and do the bridging work together. The first teams to run this evaluation will have a meaningful head start.

Request a POC

Tell us about your project. We will get back to you as soon as possible.

Thank you — we have received your request. A member of our integration team will be in touch to begin the scoping process by email.

Your information is used solely to respond to this enquiry and will not be shared.

Two Ways to Work With Us

You Choose How Much
You Want to Do

Mode 1 · Self-Service

Your Team Drives

You have an engineering team and want full control. We provide ADD Studio, the rule authoring toolchain, the scenario editor, and the full acceptance suite. Your engineers write the rules, build the scenarios, and iterate at their own pace.

Best for teams who want to own the driving policy from day one.

ADD Studio YAML Rule Authoring Full Toolchain Access
Mode 2 · Managed

Don't Want to Learn the Toolchain? That's Fine.

If you already have your own rulebook — traffic regulations, defensive driving policies, or proprietary behaviour guidelines — hand it to us directly. We translate your rules into the engine, load them, and hand back a system ready for your road test.

No onboarding required on your side. You bring the rules, we bring the engineering.

Rule Translation Managed Integration Road-Test Ready
Early Partner Programme

Be Among the First to Validate ADD
on Real-World Data

ADD has been rigorously validated in simulation — 2,262 test items, 10 core scenarios, all green. What comes next is real-world perception data, and we are actively seeking a small number of OEM and Tier-1 partners to run the first production-environment evaluations together.

Early partners get direct access to the core team, input into the roadmap, and the opportunity to shape how ADD integrates with real vehicle stacks — before it becomes a standard offering.

Apply for Early Partner Status →
About the Team

Who Is Behind ADD

ADD-AV was founded by Shaobo Qiu, who previously served as Director of Autonomous Driving at FAW (First Automobile Works) Technical Center, and as a Research Fellow at the Robotics Center of the University of Electronic Science and Technology of China (UESTC), Chengdu.

The project is developed in collaboration with the School of Vehicle and Mobility, Tsinghua University, and is under the technical guidance of Academician Li Jun of Tsinghua University, a leading authority in automotive engineering and intelligent vehicle systems in China.

FAW Technical Center UESTC Robotics Center Tsinghua University Academician Li Jun
Direct Contact

Other Ways to Reach Us

General enquiries
hello@add-av.com
Product questions, press, general information
Technical integration
engineering@add-av.com
API, perception bridge, and architecture support
LinkedIn
linkedin.com/company/add-av
Technical updates and industry news