Symmetry is one of the beauties of nature, and often the essential ingredient in an elegant engineering solution, but few vehicles are as symmetrical as they might superficially appear. Transverse powertrains and off-axis locations for drivers and driving controls, not to mention numerous other ancillaries, make most cars much less symmetrical than their sheet metal might suggest. But one area where an even side-to-side balance has ruled for a long time - though perhaps not for much longer - is in the distribution of tractive and braking torques between the wheels.

In a corner, the wheels on the side of the car closest to the apex must follow a shorter path than the outer wheels, and so must rotate more slowly. A conventional open differential allows the inner drive wheel to rotate more slowly than the outer while still transmitting drive torque equally to both wheels on the axle. Most of the time that works well, but because the open differential delivers equal output torque to both wheels, only one wheel has to lose grip for the output torque of the whole differential to drop to almost nothing. That can easily occur in corners, when weight is transferred from the inner wheels to the outer, and the inner wheel tries to spin. In straight-line acceleration it can occur where the road material has a variable coefficient of friction or an uneven surface.

Limited-slip differentials (LSDs) first became popular in the US in the 1950s, as engine capacities rapidly increased and automatic transmissions with torque convertors became popular. As a result, tractive effort available at the rear wheels raced ahead of improvements in tyre technology. Early LSDs commonly used clutch packs between the output bevel gears and the differential cage, which came into operation when high input torque tried to force the bevel gears of the two shafts apart. Any difference in output shaft speed - caused as one wheel started to spin - was resisted by the friction of the clutches on the differential cage.

Torque biasing differentials such as the Torsen, Eaton and Quaife designs have achieved similar ends through a different approach. These differentials use a helical geartrain with carefully chosen tooth forms which resist rotation in a given direction: a spinning wheel tries to drive the helical gear against its primary direction of rotation, and the resistance of the gear to rotation limits the wheelspin. Torque is transferred to the opposite wheel, in a ratio of anything up to 5 to 1, provided both wheels have at least some grip. Torque biasing differentials started to appear in the 1980s, but by then the advent of robust electronic systems was popularising another form of asymmetrical control.

This time, though, it was brake force which was being controlled at individual wheels. Anti-lock braking arrived in 1978 - if you ignore previous low-volume adaptations of Dunlop's Maxaret system, designed for aircraft - when Bosch ABS became available in its first production application, on the Mercedes-Benz S-class. Through sophisticated electro-hydraulic control of the brake line pressure to each individual wheel, and wheel speed sensing at all four corners, the Bosch system could detect and largely avoid wheel lock under heavy braking. Often this resulted in shorter stopping distances in emergencies - particularly when the tyre/road friction coefficient was low, for instance when the road was wet. Maximising the retardation at the wheels with the least grip also reduced the difference in braking force on each side of the car in many situations, minimising yaw, though smaller cars need extra control circuitry and control logic to avoid stability problems when braking on 'split-mu' surfaces combining high and low friction areas.

Once anti-lock braking had become established, the sensor hardware was quickly put to other uses. Electronic traction control could easily be implemented to throttle back the engine and apply a single brake to slow a spinning drive wheel, at the same time raising the torque output of the (open) differential so that the other drive wheel could transmit useful tractive effort. The price paid was in brake wear and fuel consumption.

The next step along the same road deliberately used the braking system to yaw the vehicle, as a corrective measure. It implicity assumes that drivers are largely capable of choosing and following an appropriate path but are less effective at rapid correction of unexpected changes in yaw rate - a reasonable assumption in most cases. Electronic stability control (ESC) applied individual wheel brakes to correct the attitude of the car when significant front or rear wheel slip is detected. The most sophisticated systems took inputs from the ABS wheel speed sensors, measured the angle and rate of turn of the steering wheel, and were also provided with yaw rate and lateral acceleration sensors. So successful has ESC been in improving safety that EuroNCAP recently called for it to be standard on all new cars, though it has its limitations: brake events inevitably slow the car down, and during acceleration an individual brake may not be able to generate the force required for an appropriate course correction.

Neither of these problems occurs with electronically-controlled redistribution of drive torque. So-called 'vectored torque' systems can transfer torque to the outer wheel to increase a vehicle's yaw rate and reduce understeer. Similarly, torque can be transferred to the inner wheel to reduce yaw rate and avoid oversteer. In very low grip situations such as ice and wet grass, vectored torque systems can also improve traction by sending torque to the wheel which can most usefully use it.

Over the next few years a growing range of OEMs will adopt this 'vectored torque' technology. So far, production applications have been largely limited to four-wheel drive performance models from two Japanese OEMs - Honda and Mitsubishi - while Audi and BMW are both about to offer torque vectoring systems on their four-wheel drive vehicles. Four-wheel drive cars often suffer from power understeer, as the tractive effort going through the front wheels reduces the maximum cornering force achievable through the front tyres. Torque vectoring systems can achieve the responsiveness of rear-wheel drive without sacrificing all-wheel drive traction, and research has shown that only one of the two drive axles needs torque transfer capability to produce effective results.

Several vectored torque systems have been revealed, broadly following the same outline. Alongside the differential is a torque transfer geartrain which can be over- or under-driven under the control of a pair of clutch packs. Each clutch controls the transfer of torque in one direction across the axle.

Since 1996 Mitsubishi has fitted its own vectored torque system, known as Active Yaw Control (AYC), to a variety of models - notably several generations of the high-performance Evolution saloon. The latest version of its system - Super-AYC, fitted to the Evo X - uses a planetary geartrain in place of conventional differential bevel gears, and a torque transfer geartrain controlled by twin hydraulically-operated. S-AYC can transfer up to 10% more torque between the rear wheels than the previous system.

In addition to inputs from the ABS wheel speed sensors and engine management system, S-AYC also measures steering angle, longitudinal and lateral acceleration, and brake line pressure. As with other systems, yaw rate and steering angle are the key data: the control system compares actual vehicle attitude with the behaviour it believes the driver is aiming to achieve, and if necessary it transfers torque between the rear wheels to adjust the vehicle attitude.

In addition to its role of transferring torque between the rear wheels, the latest AYC system also incorporates a brake force control function offering an additional strategy for regulating yaw movement, which Mitsubishi says is particularly useful in high lateral-g manoeuvres. The system can apply a braking force to the inside or outside wheels to help the car maintain its line.

In a long, fast corner an Evo with the S-AYC active yaw control axle will initially understeer mildly, but unlike many four-wheel drive cars which understeer more and more as power is applied the Evo holds its track as the axle directs more torque to the outer rear wheel to increase the yaw rate. The result is secure, but responsive, handling which flatters the driver without the intrusive intervention of a stability control system.

Honda's more recent SH-AWD system works on similar lines, as does the Ricardo's 'Torque Vectoring' system revealed in 2006. Ricardo uses a pair of 200Nm wet clutch packs on one side of the final drive to control a torque transfer gearset, typically allowing a variation of wheel speed from side to side of 20% and cross-axle torque transfer of up to 1400Nm with 90 per cent of demanded torque available just 100ms after request.

The advantage of the Ricardo system seems to be its compact size and light weight. It fits within the package envelope of an existing differential, so few changes to housings, vehicle structure and suspension geometry are required, cutting costs and minimising development time.

Ricardo has demonstrated Torque Vectoring on an Audi A6, and several OEMs are said to be evaluating it for a variety of different applications. Front, rear and all-wheel drive configurations are being studied, with crownwheel torques ranging from 3000Nm to 5000Nm and annual production volumes from 5000 to 100,000 units. Similar technology can also deliver interesting manoeuvrability advantages for multi-wheeled military vehicles: yaw rate control can be used to improve their turning circle, and even allow them to 'spin on the spot' like a tracked vehicle. Production vehicles fitted with the Ricardo system are expected by 2011. The company is also in discussions with potential manufacturing partners.

Vector-Drive is another version of the vectored torque idea developed by ZF and GKN, and available in the new BMW X6. BMW calls it 'Dynamic Performance Control'.

Though it uses similar principles, the ZF/GKN design is based around modular components which should be adaptable to a range of vehicles, which may result in a lower cost per unit, though it may prove to be bulkier and heavier than some competing technologies. The Vector-Drive axle adds a planetary gear set, an electronically controlled multi-plate wet clutch and a two-stage ball-ramp mechanism on each side of the axle. Up to 1800Nm of torque can be transferred between the wheels, and the current torque split can be viewed by the driver on a dashboard display. An interesting advantage of this system is that it is not limited to operation under power - it continues to operate in over-run situations, to improve stability.

Dynamic Performance Control is operated by BMW's Integrated Chassis Management system, a high-performance electronic control network using high-speed FlexRay data transmission technology which also manages the four-wheel drive system, stability control, active steering and adaptive ride systems.

The challenge is to engineer these systems so that they protect all drivers from their own mistakes while still allowing skilled drivers to make the most of the vehicle's dynamic envelope. This is largely a matter of appropriate control software, and demands the recognition from developers that drivers have different levels of interest and commitment in their use of vehicles. In many market sectors it will be acceptable to engineer all these systems for optimum safety at the expense of responsive handling, but particularly in premium sectors it will be important to build in driver-selectable modes giving control over the level of authority the systems have. Ultimately, and sadly for enthusiastic drivers, the level of control available may be determined not by engineers but by product liability lawyers.

Reduce vehicle dynamics to its simplest and you end up with a famous, if rather pessimistic, old aphorism which says that stable, understeering cars go off the road forwards, while unstable, oversteering cars go off the road backwards. If suitably implemented, yaw control made possible by asymmetric brake and drive torques take us a step closer to the ideal middle ground - keeping the vehicle on the driver's chosen line in all circumstances.

Published in European Automotive Design 2008