Brakebetter

Brake Robot

In this instalment, I will examine the topic of automated driving, and look at how the rise of autonomous vehicles will change the demands placed on brake systems

The first thing we need to establish is what constitutes an autonomous car, then look at where we are today, what can we expect in the next decade, and a look further into the future. We will discuss how driving habits will adapt, and how transport solutions that we have known for a long time may start to change.

Then we will examine some of the technical demands that will change what brakes are built into vehicles, talk about how brakes in the future will be used and maintained, and what current technical demands will cease to have significant relevance, and come up with a forward looking view of the braking requirements of our (bright and shiny) autonomous future.

Level Up

Let’s start with what is an autonomous vehicle. The most useful and widely understood reference is the SAE Autonomous Vehicle scale (J3016), which looks at the driver’s tasks (and related vehicle ability), and categorises these in five levels.

LevelNameCommentControlsDriverFallback
0No AutomationHumanHumanHuman
1Driver Assistance“hands on”ComboHumanHuman
2Partial Automation“hands off”SystemHumanHuman
3Conditional Automation“eyes off”SystemSystemHuman
4High Automation“mind off”SystemSystemSystem
5Full Automation“no driver”SystemSystemSystem

Level 0 is the default for today’s vehicles – and incorporates systems which may monitor road conditions and warn a driver, but don’t directly intervene in vehicle controls. Level 1 and L2 require a driver to oversee the actions of the system, and intervene if the system makes a poor decision. Such systems are becoming commonplace in new vehicles, and are gaining widespread acceptance due to the improvements in vehicle safety they provide. L3 systems promise to handle more tasks more often, and should allow the driver to engage in other tasks. L3 systems can require human intervention, but with a small buffer (typically 30s). The primary difference between L4 and L5 is that L4 is expected to be fully autonomous, but within limited road conditions – for example on motorways and national roads, and requiring some human assistance at more complex environments. L5 systems are vehicles which drive themselves, with no provision needed for human drivers. Since 2018, a fully autonomous taxi service has been operating in Arizona, and many more are planned. Meanwhile, in Sweden, fully autonomous trucks are entering public testing.

T-pod.jpg
By LinneakornehedOwn work, CC BY-SA 4.0, Link

Brakes for robots

Now that we have a handle on what constitutes an autonomous vehicle, and a view of where such vehicles are in development, it gives context to discuss what sort of brakes might be required. The first significant change (to enable even L1 abilities) is on the actuation side – the ability to engage the braking system without driver input. While this problem was originally addressed in the vacuum booster, it is now more common to use the ESC system to generate brake pressure as required. In such a system, while the “robot” can generate braking force via the brake assistance, if there is a failure of the robot or the assistance mechanism, the fall-back is the driver’s unassisted foot. In this scenario, the addition of autonomous braking doesn’t necessarily create extra redundancy complexity – the sizing of the hydraulic components will be as before, and the driver is required to deploy a similar amount of force to achieve the braking power.

Safer than humans

Dalek
Not all robots are safe road users – image used under CC licence

However, to get the best possible performance from such systems, it is worth fitting actuation systems which minimise braking distances, and therefore reduce the severity of any accidents. The term Time-to-Lock (TTL) is used as a dominant metric here – a measure of how long it takes for the brake system to achieve wheel lock. Typically, TTL below 200mS are achievable with electromechanical actuation systems, and pressure rise rates in excess of 200Bar/s are necessary. This performance is greater than a human could reliably achieve, and in terms of accident outcomes, such fast acting brake systems represent a significant improvement in survival rates. It is this safety performance that has accelerated the adoption of Autonomous Emergency Braking – so much so that the EU will mandate it on all new vehicles from 2022.

So, the first new requirement we can see for autonomous vehicles is fast-acting braking systems, which minimise braking distances. This will mean a change in energy supply systems for braking – and most likely a shift from vacuum boosters towards electromechanical boosters, and in the medium term, this may mean a shift away from hydraulics altogether. Another positive reason for such systems is the extra time freed up for environmental sensing – the quicker the brakes can react, the later the decision must be made.

Moving (slowly) towards autonomy

Herbie!
An early vision of autonomous cars – used under CC licence

The step from L2 to L3 brings with it some significant challenges at a vehicle level, and the brakes also need to step up their game here. L3 brings what’s referred to as conditional autonomy, and in practice, this means low speed autonomy, in well defined driving conditions. So traffic jams on a motorway, for example. While that might not seem the most dynamically challenging environment, the vehicle is now responsible for judging its environment, and making all decisions – effectively driving the car. The human driver is free to engage in other tasks, but should be “on call” to take back control if the system requires a bail-out.

In terms of braking, as well as the previous requirements for L2, this new level brings with it a requirement to have a braking system that can still operate autonomously for a short period of time after a fault is detected, and bring the vehicle to a safe position until the driver can get involved again. In short – the brakes must be able to tolerate a loss of energy supply for up to 30s, and then revert to a mechanical or hydraulic back-up. In practice, this can mean that the power supply for the Electronic Park Brake should be independent from the hydraulic brakes. Or separate power supply between the ESP system and an electromechanical booster. In both cases, basic brake modulation must be possible.

Another area for consideration here is driver engagement. While it is not strictly a braking topic, if a driver is been woken from whatever restful activity they were previously engaged in, thought should be given to the response characteristics of the primary controls that the driver must immediately use. A stiff pedal curve may be excellent for judging grip levels on your favourite race track, but probably not the right pedal feel for this scenario. An ability to offer high levels of boost would be useful, and make the change in vehicle response less alarming.

Level 4 – the rise of the machines

Things get really interesting at L4 – high automation. Here we are considering autonomous driving for long periods of time, over a large variety of speeds and roads, and without a driver available to catch any spills. The main difference between this and full automation is that the machine can complete a journey in some environments, but cannot cover all roads and conditions. So there is a need for a driver to complete some journeys. Handover would be possible on the move, but if the driver is unavailable, the vehicle must come to a stop in a safe place, and await outside intervention.

In many ways, this represents the highest technical challenge for braking – having all the requirements and paradigms of human interaction, but also a fully formed autonomous system to boot. As before, the system must be fault tolerant without human intervention, but in a difference to Level 3, it is expected that the vehicle can continue it’s journey after a significant fault (such as loss of energy to the braking system). This requirement may mean a specific power reserve in one of the brake actuation systems, or more likely, a dedicated redundant power system design, so that other primary safety systems (such as steering) can also be supported in such a scenario. If this design choice is taken, and the braking system can rely on a redundant power supply, it simplifies the hardware design.

However, in either case, this is a Brake-by-Wire system, and so choices for driver interaction beyond hydraulic systems exist. This can radically alter the pedal box layout, and in all likelihood the benefits for crash performance will mean hanging pedals connected to large mechanical elements in engine bays will become a thing of the past. A moveable brake pedal, which can be retracted when in autonomous mode, would have obvious comfort benefits as well as simplifying package and complexity issues. At time of writing, braking legislation still mandates a mechanical connection, but the safety balance, as well as recent developments in steer-by-wire (UN R79) may soon lead to a change here.

So, how do we expect to actuate these systems? Well, the two obvious choices are with hydraulics (powered by a central hydraulic unit, or possibly near the wheels), or electro-mechanical brakes (powered in-wheel). The adoption of by-Wire braking systems using a central hydraulic unit (with a redundant M/C) is already in the market, so it is likely that this may be the initial route taken by the industry.

Level 5 – look mom, no pedals

waymo
A fully autonomous vehicle – image used under CC licence

Ironically, the outlook for a L5 brake system will be somewhat simpler than a L4 system. Firstly, the need for any human interaction is completely removed, meaning there is no longer a pedal box to speak of. While this may seem like a minor foot-note (!), it actually has significant implications for the brake system as a whole, but most especially the foundation brakes. If the driver doesn’t have access to the accelerator, it is likely that repeated vehicle accelerations will be a thing of the past, and so brake fade requirements will simplify. If the driver doesn’t have access to the accelerator, it is likely that max vehicle speeds will reduce significantly, meaning ultimate thermal loads into the brake system will similarly reduce. If the driver doesn’t have access to the accelerator pedal, they can’t override the minimum safe following distances, brake late into junctions, take blind curves at aggressive speeds (and so on), so its likely that high energy brake events will be less likely, meaning greater reliance on regenerative brake systems. This also has significant ramifications for actuation sizing, but the ability to achieve minimal Time-to-Lock, brake distances and highest possible vehicle stability will remain.

In short, removing the accelerator pedal probably will have the single biggest effect on brake system design – as the requirements for vehicle performance will be so widely different from today’s systems.

Summing up

The rise of autonomous driving will mean the rise of increasingly complex and capable brake systems. While it is true to say there are examples of every type of vehicle already on our roads, the widespread adaption of these vehicles will be required to see significant changes to the braking world. The timeline for this remains unclear, but it is pretty obvious that the technical challenges can be solved.

For brake systems, the most compelling trend will be additional complexity of control systems, redundancy in system design, and finally simplification of controls as well as downsizing of foundation components. We are well underway with the complexity – indeed, brake control systems have been increasing exponentially since the introduction of ABS – but Autonomy will see a step change in requirements, and with it the need for significant new investments.