Safer Roads Need Systems That Act Before the Impact

Road safety is still too often treated as a problem to be reviewed after the damage is done. But on high-speed freight corridors, risk does not wait for hindsight. It builds in seconds, through fatigue, distraction, shrinking following distance, and delayed reaction. By the time a crash is recorded, the real opportunity has already passed.

A government-commissioned estimate pegged the socio-economic cost of a road fatality at around ₹91 lakh. For fleets, the impact of a major crash rises even further once downtime, claims, legal effort and missed deliveries are added. In practical terms, a single serious incident can wipe out months of operational gains. That is why road safety needs a new operating logic. It cannot remain a reporting exercise. It has to become a prevention system.

The Real Cost of Hindsight: Shifting Road Safety from Reporting to Prevention

The reality on the road is that risk forms in seconds. Under speed and time pressure, fatigue shows up before failure does. Micro-sleeps and attention loss do not announce themselves. They surface briefly, then turn into late braking, drift or poor judgment. This is exactly why prevention needs to work as a live loop. Systems must detect early cues in milliseconds and respond while the moment is still recoverable.

That is where the shift to edge intelligence becomes decisive. Road dynamics do not wait for cloud round trips. On live routes, latency and inconsistent connectivity can consume the only window that matters, which is the moment before impact. Safety systems need to process what they see in the vehicle, at the edge, and in real time. This is particularly relevant in India, where operating conditions are dynamic, road environments are uneven, and connectivity cannot always be assumed.

But speed alone is not enough, and context matters just as much. The most useful systems are the ones that can explain the “why” behind the risk, not just the “what.” It can distinguish between hard braking due to a cut-in and a distraction event." It can read following-distance changes, unstable traffic flow, drowsiness cues and shifting road conditions in context. The goal is not to generate more alerts. It is to generate the right alerts. Precision matters because nuisance alerts erode trust, and once trust goes, driver engagement tends to go with it.

Validating Technology for the Complexity of Indian Corridors

This is also why ADAS in India is moving into a more meaningful phase. For years, the conversation was centred on promise, but now it is shifting toward proof. ARAI’s Government-backed ADAS test environment in Pune signals an important move from ambition to validation. A controlled proving ground matters because it allows active safety systems to be tested in repeatable, India-relevant conditions rather than being judged only in ideal or imported scenarios.

That distinction is important. Indian roads present a complexity that cannot be reduced to lane markings and clean traffic flow. There is mixed vehicle density, variable road quality, vulnerable road users and a constant negotiation of space. A system that performs well in controlled conditions but fails under real pressure is not deployment-ready. The next stage of this conversation, therefore, is not simply about feature readiness but also about operational readiness. Can these systems hold up over long-haul routes, under glare, at night, in dense traffic and across high-mileage commercial use cases? That is the benchmark that matters.

Physical AI: Closing the Safety Loop

The economics support this shift as well. Crash-avoidance technologies are increasingly being seen not as optional features but as part of safety infrastructure. The commercial ADAS market is growing rapidly, and with good reason. Evidence for closed-loop intervention is becoming harder to ignore. When systems act before impact, outcomes change. That is the larger lesson behind the growing focus on automatic emergency braking and related technologies. The future of road safety will belong to systems that do not just observe risk but reduce it.

This is where the idea of physical AI becomes especially useful. In simple terms, physical AI is AI that perceives, reasons and acts in the physical world in real time. For mobility, that means moving beyond passive recording and towards a closed-loop safety model. Analyse the scene, estimate the risk and then trigger an intervention while the moment is still reversible.

Additionally, it also learns from the outcome. Road safety needs that control-loop mindset because risk on the road does not wait for analysis cycles. It requires systems designed to operate in the moment.

Building a Human-Centric Safety Culture

Yet technology alone will not improve outcomes at scale unless fleets also rethink the human side of safety. Telematics has already become the operating layer for many fleets. The next layer has to be behavioural. Too often, safety systems are still built around gotcha logic. They highlight what went wrong, issue repetitive alerts, and leave drivers feeling judged rather than supported. That approach may create visibility, but it rarely builds lasting engagement.

Positive driving recognition offers a better path. When systems acknowledge safe behaviour alongside risky behaviour, coaching becomes more balanced, more credible and more actionable. It feels fairer, and it earns more trust. In high-mileage, deadline-driven operations, that matters more than many leaders realise.

Drivers are far more likely to engage with a system that reflects the full picture of their performance rather than one that notices them only when something goes wrong. Recognition does not dilute accountability; it strengthens adherence by making the feedback loop more human and more durable.

That, ultimately, is the direction road safety needs to move in. Not toward more footage, more dashboards, more alerts for their own sake. It needs to move toward systems that can detect risk early, understand it accurately and act in time. Safer roads will depend on technologies that work under real driving constraints and on safety cultures that drivers can trust. The real breakthrough will not come from knowing more after a crash. It will come from preventing the crash in the first place.

Ajit Nair is the Director of Product Management at Netradyne. Views expressed are the author's personal.