Vehicle detection technology forms the perceptual foundation of any smart traffic system. Without reliable, accurate detection, adaptive signal controllers have no useful data to act on, real-time monitoring platforms cannot reflect actual road conditions, and intersection performance degrades toward fixed-time guesswork. Selecting and specifying the right detection approach requires a clear understanding of each technology's operating principles, strengths, and limitations in the context of Australian road environments.
How vehicle detection fits into the wider signal system
Modern traffic signal controllers depend on a continuous feed of presence and count data to make timing decisions. Detection equipment captures that data at the intersection and passes it upstream, either to a local controller or to a network operations centre. The quality of that data directly shapes signal performance. A poorly placed or misconfigured detector introduces latency, generates false calls, or misses vehicles entirely, all of which ripple through the network as inefficiency or unsafe gap acceptance. For sites deploying smart intersection design principles, detection accuracy is not optional: it is the enabler for everything else the system is meant to do.
Inductive loop detectors
Inductive loop detectors remain the most widely deployed vehicle detection technology across Australian road networks. A wire loop is cut into the pavement and connected to a detector unit inside the traffic cabinet. When a vehicle passes over the loop, its metallic mass disturbs the loop's inductance and triggers a detection event. Loops are well understood, easy to maintain, and relatively inexpensive per lane. Their primary weaknesses are pavement disruption during installation, susceptibility to damage from heavy vehicle movements and road resurfacing, and an inability to classify vehicle types beyond basic presence and count. For high-traffic arterials where pavement maintenance cycles are frequent, lifecycle costs can become significant.
Video detection systems
Camera-based video detection uses image processing algorithms to identify and track vehicles within a defined field of view. A single camera can cover multiple lanes simultaneously, reducing the number of physical installations needed at complex intersections. Modern video detection systems classify vehicles by type, measure queue lengths, estimate speeds, and feed that richer dataset into adaptive control algorithms. Performance degrades in low-visibility conditions including fog, heavy rain, and night-time glare, so video detection is typically deployed alongside complementary sensors rather than as a standalone solution. Processing can occur on-board the camera unit (edge processing) or be offloaded to a central server, with the former offering lower latency and greater resilience when network connectivity is intermittent.
Radar and microwave detection
Radar-based detectors emit microwave signals and measure the reflected return to determine vehicle presence, speed, and in some configurations, vehicle type. Unlike video systems, radar performance is not affected by lighting conditions or most weather events, making it well suited to sites with high ambient glare, such as motorway ramps, tunnel portals, or coastal roads subject to sea fog. Side-fire radar units mounted on roadside poles can monitor multiple lanes across a wide detection zone. Forward-facing units provide point speed measurements and queue detection with high precision. Radar is increasingly specified on arterial corridors where all-weather reliability is a contractual requirement.
Thermal and infrared sensors
Thermal imaging detectors operate on heat signatures rather than visible light, making them largely immune to solar glare and effective in complete darkness. They are most commonly deployed at pedestrian crossings and intersections where reliable detection of slow-moving or stationary objects matters. Thermal sensors do not capture image detail usable for enforcement or identification purposes, which simplifies privacy compliance, but they also cannot read licence plates or support secondary applications such as incident detection. For specialised environments including school zones, hospital precincts, and heavy pedestrian corridors, thermal detection offers a compelling reliability case.
LiDAR and 3D sensing
LiDAR (light detection and ranging) emits laser pulses and builds a three-dimensional point cloud of the detection zone. This allows precise vehicle classification, including distinguishing motorcycles from passenger cars from heavy vehicles, measuring vehicle dimensions, and tracking trajectories through complex manoeuvres such as turns and merges. LiDAR is increasingly specified at high-value smart intersections where granular data feeds are needed to support AI-driven signal optimisation. The technology remains more expensive per unit than radar or video, but costs have fallen considerably over recent years as automotive-grade LiDAR components have reached volume production. Sites deploying AI-driven traffic signal control benefit most from LiDAR's rich classification output, as the additional data dimensions improve model accuracy and reduce false positives in the control algorithm.
Multi-sensor fusion and data integration
No single detector technology is optimal across every site condition, traffic type, and performance requirement. Best practice for complex or high-volume intersections is multi-sensor fusion: combining outputs from two or more detection technologies so that each compensates for the other's weaknesses. A typical configuration might pair video detection for queue measurement and vehicle classification with radar for all-weather speed and presence confirmation. The fused data stream feeds the controller with higher confidence readings than either sensor alone could provide. Fusion processing can be handled at the roadside cabinet using embedded edge hardware or centrally via a traffic management platform. Either way, the architecture needs to account for sensor synchronisation, data format compatibility, and failure mode behaviour when one input drops offline.
Key specification considerations
When specifying vehicle detection for a new installation or upgrade, several factors consistently influence the final technology selection:
- Site geometry: lane widths, approach angles, presence of medians, and overhead structure availability all constrain mounting options and detection coverage zones.
- Environmental conditions: sites subject to regular fog, heavy rain, or direct sun exposure need detection technologies with demonstrated all-weather performance, not just nominal specifications.
- Traffic composition: high proportions of motorcycles, cyclists, or heavy vehicles require detectors with strong classification capability to generate accurate data for the controller.
- Integration requirements: the detector's output protocol needs to be compatible with the signal controller and any upstream traffic management platform in use.
- Maintenance access: pavement-embedded technologies require road closures for repair; above-ground sensors are generally more accessible but need periodic cleaning and alignment checks.
- Whole-of-life cost: procurement price is rarely the dominant cost driver. Installation complexity, maintenance frequency, and the consequence of detection failures all contribute substantially to lifecycle expenditure.
Standards and compliance context
Vehicle detection equipment deployed on Australian public roads must comply with relevant Austroads technical guidelines and applicable state transport authority specifications. Detection performance benchmarks, including minimum detection probability and maximum false call rates, are typically set out in project-level technical requirements documents. Installers and suppliers should confirm that equipment meets or exceeds these benchmarks under the site's worst-case environmental conditions, not just under controlled test conditions. Cabinet integration, including wiring, connector types, and detector unit form factors, should also be verified against the controller hardware already installed or planned for the site.
Vehicle detection is not a commodity purchase. The technology chosen at the detector level determines the quality of every data-driven decision the signal system makes above it. A rigorous, site-specific selection process, grounded in real performance data rather than specification sheets, is the most reliable path to a system that performs as designed over its full service life.