Torque vectoring is the deliberate distribution of unequal drive or braking torque between individual wheels so that the difference itself creates a yaw moment that helps the vehicle turn, hold its line or recover stability. In an EV drivetrain, it is especially effective because electric motors can change torque in a few milliseconds, in both directions, and with precise control that a combustion engine and mechanical differential cannot match.
The concept matters wherever handling, traction and safety meet. A car that turns partly because the outer wheel pushes harder than the inner wheel needs less steering angle, scrubs less speed in corners and stays more predictable on low-grip surfaces. Torque vectoring is closely related to dual-motor AWD, e-axle design and regenerative braking, but it describes the control strategy, not a specific piece of hardware.
How torque vectoring creates a turning moment
Torque vectoring works by producing a longitudinal force difference across an axle. When the outer rear wheel delivers more forward force than the inner rear wheel, the force pair acts around the vehicle’s center of gravity and generates a yaw moment in the direction of the turn. Reversing the difference produces a stabilizing moment that counters oversteer.
The size of the moment is roughly the force difference multiplied by half the track width. On a car with a 1.6 m track, a difference of 2,000 N between the left and right wheels yields about 1,600 Nm of yaw moment, which is enough to change the cornering behavior noticeably at road speeds. The available effect is always limited by tire grip, because a wheel that is already near its friction limit cannot accept more longitudinal force without losing lateral grip.
A control unit decides the split many times per second. It compares the driver’s intended path, estimated from steering angle and speed, with the measured yaw rate and lateral acceleration from the inertial sensors, then commands each motor or brake accordingly. The following table summarizes the main inputs used by a typical controller.
| Signal | Source | Role in the decision |
|---|---|---|
| Steering wheel angle | Steering angle sensor | Estimates the intended path |
| Yaw rate | Inertial measurement unit | Measures the actual rotation |
| Lateral acceleration | Inertial measurement unit | Indicates how close the car is to the grip limit |
| Wheel speeds | ABS wheel sensors | Detects slip at each wheel |
| Accelerator and brake input | Pedal sensors | Sets total torque demand |
| Motor torque and speed | Traction inverter | Confirms delivered torque |
Hardware architectures that enable torque vectoring
The capability of torque vectoring depends on how many independently controlled torque sources the drivetrain has. A single motor driving an open differential cannot vector torque by itself and must rely on the friction brakes to slow one wheel. Two motors on one axle, or a motor per wheel, allow genuine left-right torque control without wasting energy as heat.
The table below compares the common layouts found in electric vehicles today and shows why the most capable systems are also the most expensive.
| Layout | Vectoring method | Response and capability | Main drawback |
|---|---|---|---|
| Single motor, open differential | Brake-based vectoring | Moderate, only reduces torque | Brake wear and energy loss |
| Dual-motor AWD, one per axle | Front-rear split plus brakes | Good front-rear balance | Limited left-right control |
| Single motor with clutch-pack differential | Clutches shift torque across axle | Good, can over-speed outer wheel | Mechanical complexity and heat |
| Twin motors on one axle | Independent left and right torque | Very high, fully active | Cost, packaging and weight |
| Four motors or in-wheel motors | Independent torque at every wheel | Highest, including counter-rotation in some designs | Unsprung mass and cost |
Twin-motor rear axles are the most common performance solution because they keep motors inboard, avoid the unsprung mass penalty of an in-wheel motor and still provide full left-right authority. Some designs can even apply regenerative braking on the inner wheel while driving the outer wheel, which recovers energy while generating a yaw moment.
Torque vectoring compared with related stability functions
Torque vectoring is often confused with traction control and electronic stability control, yet each function has a different goal and intervention style. Traction control limits wheel slip, stability control prevents a skid once it starts, and torque vectoring shapes the handling continuously before the car reaches its limit.
| Function | Primary goal | Typical action | When it acts |
|---|---|---|---|
| Torque vectoring | Improve agility and stability | Redistributes torque between wheels | Continuously, well below the limit |
| Traction control | Prevent wheel spin | Reduces torque at a slipping wheel | When slip exceeds a threshold |
| Electronic stability control | Prevent loss of control | Brakes individual wheels, cuts power | Near or beyond the grip limit |
| Limited-slip differential | Keep traction on split grip | Locks torque across an axle passively | When wheel speeds diverge |
In modern electric vehicles these functions share sensors and often share software. The practical distinction is that a well-tuned torque vectoring system reduces how often stability control must intervene, because the car stays closer to the driver’s intended path from the start.
Control strategies and tuning choices
Engineers tune torque vectoring according to the character the manufacturer wants and the conditions the car will face. A sporty calibration adds yaw on corner entry to reduce understeer, while a comfort calibration focuses on straight-line stability and smooth behavior on wet or icy roads.
| Driving situation | Torque vectoring strategy | Intended result |
|---|---|---|
| Corner entry on dry roads | More torque to outer rear wheel | Reduced understeer, sharper turn-in |
| Mid-corner at high speed | Small corrective adjustments | Stable, predictable line |
| Snow or ice | Gentle, early stabilizing torque | Less sliding, smoother progress |
| Split-friction surface | More torque to high-grip side | Straight acceleration without pulling |
| Track or sport mode | Higher yaw gain, later intervention | More agile, driver-led handling |
Most production systems use a reference yaw model, often based on a linear single-track vehicle model, to calculate the desired yaw rate. A feedback controller then corrects the error, while a feedforward term reacts to steering input before the error appears. The calibration must respect motor limits, battery power limits and the thermal state of the traction inverter.
Limitations, trade-offs and common misconceptions
Torque vectoring cannot create grip that the tires do not have. It only redistributes forces within the friction available at each contact patch, so on very slippery surfaces its effect becomes small and stability control takes over. Drivers who expect it to defeat physics may enter corners too fast.
- Brake-based vectoring wastes energy and heats the brakes, which limits it during long, spirited driving.
- Aggressive yaw gain feels artificial to some drivers and can mask the natural warning signs of understeer.
- Sensor faults or poor calibration can produce unexpected moments, so systems include fallback modes that revert to a neutral split.
- More motors add cost, mass and complexity, which is why many mainstream electric vehicles offer only front-rear control.
- It does not replace good suspension geometry; a poorly balanced chassis still behaves poorly near the limit.
A frequent misconception is that every dual-motor AWD car has full torque vectoring. In reality, one motor per axle offers front-rear distribution only, and any left-right effect still comes from braking individual wheels.
In everyday language
Torque vectoring means the car pushes harder with one wheel than another to help itself turn. Think of a rowing boat: if the rower on the left pulls harder than the rower on the right, the boat turns right without any rudder. A car does the same thing with its wheels.
Electric cars are very good at this because their motors react almost instantly and can be controlled one by one. The result is a car that corners more easily, feels steadier on slippery roads and needs less steering, but it still cannot add grip that the tires do not have.
Seen in the real world
A performance electric sedan with two motors on the rear axle and one on the front axle approaches a tight mountain hairpin at 60 km/h. As the driver turns in, the controller sends about 30 percent more torque to the outer rear wheel and applies light regenerative braking on the inner rear wheel. The car rotates willingly, and the driver needs noticeably less steering angle than in the same car with the feature disabled.
Later the same day, the road becomes wet and one side has painted lines with lower grip. When the driver accelerates out of a junction, the system shifts torque toward the wheels on the higher-grip asphalt, so the car moves forward in a straight line instead of pulling to one side. The outcome is a calmer, safer drive with fewer stability control interventions.