Traditional urban traffic management treats infrastructure and vehicles as completely decoupled systems. Traffic signals operate reactively, adjusting their phases based on localized loop detectors or cameras that count vehicles waiting at an intersection. Meanwhile, drivers navigate reactively, accelerating until they encounter a red light and then braking to a halt.
This decoupled approach creates massive systemic inefficiencies. Every time a vehicle—especially a heavy-duty truck or a tightly packed platoon—comes to a complete stop, it discards its kinetic energy as heat. Accelerating back to cruising speed requires a massive expenditure of fuel and introduces stop-and-go shockwaves that ripple backward through city streets.
The core insight of integrated routing and signal control is to break down this wall by establishing a bidirectional feedback loop between infrastructure and vehicles. By co-optimizing traffic signal timing and vehicle trajectories simultaneously, the network transitions from a reactive system to a proactive, synchronized choreography.
The Physics of Trajectory Smoothing
To understand why this approach works, consider an individual vehicle approaching a signalized intersection. In an isolated system, the driver only knows the current state of the light. If the light is red, they maintain speed until they must brake.
In a co-optimized system, the infrastructure broadcasts its future Signal Phase and Timing (SPaT) data to the vehicle via vehicle-to-everything (V2X) communication. Armed with this forecast, the vehicle can perform trajectory smoothing. If the vehicle knows a light will turn green in exactly fifteen seconds, it can calculate an optimal deceleration or acceleration curve to glide through the intersection precisely as the light changes, never dropping its speed to zero.
As demonstrated by Hsu et al. in real-world field trials with heavy-duty trucks, preserving this momentum dramatically reduces fuel consumption because it avoids the energy-intensive process of launching a stationary load. Salvato et al. show that when an infrastructure controller broadcasts scheduled signal changes in advance, connected vehicles can solve local optimization problems that minimize energy consumption without sacrificing intersection throughput. Friedman et al. formalize this concept through "Ride-the-Green-Wave" systems, which provide real-time speed recommendations to drivers to ensure uninterrupted maximum flow along arterial corridors.
The Handshake of Joint Optimization
While providing speed advice to a single vehicle is beneficial, true network efficiency requires a simultaneous mathematical handshake between the signal controller and all approaching vehicles.
When optimized in isolation, a traffic signal might extend a green phase to clear a visible queue, unaware that a large cluster of vehicles is just out of range. Conversely, if the signal timing is fixed, vehicles are forced to adapt their speeds to sub-optimal windows. As noted by Han et al., optimizing signal timing to accommodate transit schedules without integrating vehicle trajectory control severely limits overall efficiency.
+--------------------------------------------------------+
| Joint Optimization Loop |
+--------------------------------------------------------+
| |
| +-----------------------+ Predicted Trajectories |
| | Vehicle Trajectory |------------------------+ |
| | Planning Module | | |
| +-----------------------+ v |
| ^ +---------------+
| | | Traffic Light |
| | Speed Guidance / | Control Logic |
| | Signal Phase Forecast +---------------+
| +------------------------| |
| +---------------+
+--------------------------------------------------------+
To solve this, researchers formulate the problem as a joint optimization framework. Busch et al. utilize deep reinforcement learning to jointly control traffic lights and provide real-time speed advice, demonstrating that coordinating these two interdependent variables minimizes idle time far better than optimizing signals alone.
Mathematically, this is often structured as a two-part optimization problem. As detailed by Xu et al., the system first solves a scheduling problem to determine the optimal crossing sequences and arrival times for vehicles at an intersection. Second, it solves an optimal control problem to determine the precise speed and acceleration inputs required for each vehicle to meet that schedule. Dong et al. reinforce this by showing that simultaneously optimizing individual speed guidance and traffic signal phase offsets maximizes overall network throughput.
Platoon Synchronization and Collective Clearance
The benefits of co-optimization scale dramatically when applied to vehicle platoons. A platoon of connected and automated vehicles (CAVs) can act as a single, coordinated macro-entity.
Smith et al. demonstrate that by integrating infrastructure-to-vehicle data, a platoon leader can make collective decisions about intersection clearance. If the leader determines the entire platoon cannot safely clear the intersection before a signal change, it modulates the approach speed of the entire group early, preventing a dangerous and inefficient chain reaction of hard braking.
Crucially, co-optimization enables counter-intuitive emergent behaviors. Jayawardana et al. show that an autonomous vehicle can improve group efficiency by strategically slowing down a lead vehicle. By intentionally delaying the platoon, the lead vehicle synchronizes the entire following group with the upcoming green phase, allowing a massive block of traffic to pass through the intersection without a single vehicle tapping its brakes. To handle highly congested scenarios, Yu et al. integrate trajectory planning with dynamic virtual waiting zones within the intersection, ensuring smooth passage for automated vehicles even under oversaturated conditions.
Proactive Path Clearing for Emergency Vehicles
The most high-stakes application of this paradigm is emergency vehicle (EMV) routing. Traditional emergency preemption systems use a simple "green wave" mechanism: when an ambulance is detected, the upcoming light turns green. However, this sudden change often forces civilian traffic to scramble, creating localized bottlenecks and gridlock that ultimately block the ambulance anyway.
Co-optimizing routing and signal control solves this by creating a proactive, bidirectional feedback loop. Ding et al. propose a system where the emergency vehicle's routing module updates its path based on current signal states, while the signal control module simultaneously adjusts phases along that path to clear out civilian traffic before the ambulance arrives.
In Cheng et al., the authors expand this into a role-aware multi-agent reinforcement learning framework. The system coordinates traffic lights, emergency vehicles, and regular vehicles simultaneously. By dynamically adjusting signals and steering civilian traffic away from future emergency paths, the framework enables emergency vehicles to bypass bottlenecks entirely without causing a chaotic ripple effect across the rest of the city. Even in near-term deployment scenarios with low market penetration, Suo et al. demonstrate that co-optimizing the maneuvers of even a single connected vehicle with adaptive traffic signaling can significantly accelerate emergency corridor clearance.
Scaling to the Network Level
When applied across an entire urban grid, this integration extends beyond individual intersections to macro-level traffic management. For example, Guo et al. demonstrate a proactive strategy that synchronizes variable message signs (VMS) with traffic signal control, using information routing to influence driver demand, which in turn optimizes signal timing across a divergent network. Similarly, Hu et al. show that simultaneously managing interconnected traffic systems, such as ramp metering and perimeter control, yields superior results compared to optimizing each component in isolation.
Ultimately, by treating infrastructure and vehicles as a single, tightly coupled computational system, cities can maximize throughput, slash emissions, and drastically reduce travel delays without laying a single new lane of asphalt.
Key Foundational Work
Improving Urban Traffic Throughput with Vehicle Platooning: Theory and Experiments (2020) — contributed 2 concept elements to this direction.
Early Findings from Field Trials of Heavy-Duty Truck Connected Eco-Driving System (2019) — contributed 2 concept elements to this direction.
Sequential Optimization of Speed, Thermal Load, and Power Split in Connected HEVs (2019) — contributed 2 concept elements to this direction.