WeRide, Grab to test full autonomous shuttle service in SG

Singapore’s Land Transport Authority has approved WeRide and Grab to test their full Ai.R autonomous vehicle fleet in the Punggol district.

The partners began AV testing in October 2025 and plan to increase test runs on shuttle service routes by up to four times by year-end.

The Ai.R fleet includes 10 GXRs and one Robobus, serving two routes that connect residents to MRT stations and malls.

Safety operators will be on board during testing and early public rides, with the first group of trainees already undergoing on-road training.

The service aims to take its first public passengers by early 2026.

🔗 Source: WeRide

🧠 Food for thought

Implications, context, and why it matters.

The approval covers trials, not the driverless service WeRide and Grab want. Safety operators stay in place until AVs meet TR 68, Singapore’s technical reference for AV safety and cybersecurity 1. The standard covers system-level safety, Operational Design Domain (ODD) applicability, and a cybersecurity assessment framework 2.
Part 2 of TR 68 requires developers, system integrators, and operators to be competent organizations with a quality management system 1. It also calls for trained staff. That sets up a staged assurance process that runs past initial test approval.
Early-2026 public service timeline covers supervised rides with safety operators. Fully driverless service needs further milestones under TR 68 1.

Part 3 of TR 68 sets out cybersecurity test methods and attack surface reviews 2. Singapore’s Personal Data Protection Act (PDPA) also requires proper handling of in‑vehicle recordings 3. That drives demand for cloud and security tools for PDPA‑compliant logging, retention, and redaction for AV fleets.
The standard says AV security will evolve fast and that updates are planned 2. Operators will need ongoing monitoring and patching. Managed security service providers (third-party firms that monitor and manage cybersecurity) can sell continuous compliance toolkits instead of one‑off projects.
New guidance on machine learning application and software update management in TR 68 helps machine learning operations (MLOps) vendors 4. They can build platforms that track model versions, run over‑the‑air (wireless) updates, and keep audit trails that meet AV deployment rules.