AI Traffic Pilot SARTHI Flags 9,000 Violations in First Month
The Greater Visakhapatnam Municipal Corporation (GVMC) and city police have reported early success with their AI traffic management system, SARTHI (Systematic Augmented Radial Traffic & Hoop Induction).
SARTHI launched on September 10, 2025. It is active at seven busy junctions: Urvasi, Akkayyapalem, Gurudwara, Sampath Vinayaka Temple, Poorna Market, RTC Complex, and the Airport junction.
In its first month, SARTHI recorded over 9,000 traffic violations. Helmet-less riders topped the list with 3,500 cases. Other common violations were 2,300 passengers without seat belts, 1,200 red light violations, 1,050 speeding incidents, 700 illegal parking cases, and 250 vehicles moving against traffic.
The system uses Automatic Number Plate Recognition (ANPR), Red Light Violation Detection (RLVD), and Facial Recognition Systems (FRS). It operates in 12-hour shifts with help from five technology partners: Brihaspathi Technologies (Hyderabad), Meta Fusion Technologies (Noida), Data Corp (Bengaluru), NEC (Noida), and CMS Computers Limited (Mumbai).
SARTHI can issue challans automatically, monitor traffic in real time, and sync traffic signals. Officials say this reduces congestion, lowers emissions, and improves road safety.
GVMC Commissioner Ketan Garg said, “SARTHI aims to improve traffic flow and road safety. Real-time AI monitoring should reduce violations and accidents at busy junctions.”
Experts welcomed the initiative but said technology alone cannot solve traffic problems. Traffic consultant R. Prasad said, “Monitoring and integration with existing systems are essential to improve mobility.” Urban planning specialist Dr. Anitha Reddy added, “Driver education and community participation are important. Enforcement must work with awareness campaigns.”
The one-month pilot will end with a review. Authorities plan to expand SARTHI to 50 of Visakhapatnam’s 115 major intersections. With traffic increasing, city officials see SARTHI as a key tool for safer, smoother roads.