如果要为一辆传统汽车拟定变速箱和传动系统的规格，有时可以遵循一些设计套路，但如果对象换成自动驾驶汽车，那就毫无“套路”可言了，汽车咨询公司主管Lee Sykes如此表示。

作为传动系统工程咨询公司DSD（驱动系统设计）的商务总监，Sykes表示：“在设计自动驾驶汽车时，我们可能需要采取截然不同的方式。为了找到合适的解决方案，我们应透彻理解从SAE L1到L5的每一个过渡过程及其对功能优先级的影响。我们必须满足不同过渡阶段的不同要求。”多年来，DSD一直专注于未来变速箱和传动系统的工程设计、开发、测试和控制技术，近期在密歇根州成立了一个新的测试中心，希望帮助汽车制造商及其供应商提高内燃机车、电动车、混动车的传动系统的效率。

Skykes表示，到底多少比例的架构是沿用传统，多少是全新设计，这取决于目标级别是L5还是过渡级。他也补充道，完全自动驾驶汽车的设计布局正在变得越来越清晰。L5级系统不会被遗留系统的兼容问题所束缚，可以实现完全优化。

变速器和传动系统的规范对多项关键性能指标做出了规定，其中包括封装、耐久性、重量、效率、NVH、性能、安全性、舒适度和成本。这些指标决定了变速箱和传动系统的设计方向。不过，对于无人驾驶汽车而言，一些原本为驾驶员而设立的指标也许可以从此忽略不计了，或者至少，设计的优先顺序将会发生显著变化。

自动驾驶传动系统的套件配置将在很大程度上取决于座舱的空间要求。举个极端的例子，如果座舱采用“滑板型”底盘，最合适的解决方案是搭载轮毂电机，每个电机都采用整体减速齿轮传动。套件尺寸的设计则不可避免地取决于系统的耐久性和负载循环。

Sykes表示，如果一辆自动驾驶出租车提供全天候随叫随到服务，则会积累很高的里程数，显著增加负载循环。但从另一方面来看，没有人类驾驶员后，车辆不用再遭受人为的机械过载，也不必时不时迎合需求展现最佳性能，这会在一定程度上减少负载循环。当然还有一个福利，不会再有人因为车技被乘客责骂。

NVH的控制和调校一直是变速器和传动系统工程师面临的一大问题，而对于电动网联自动驾驶汽车来说，除了要营造安静的座舱环境以外，还有其它的挑战要应对。“如果加减速不平顺，人们一般认为是驾驶员的错。所以，对于自动驾驶汽车，大家期待的是完美平顺的加减速。预计未来的NVH调校将包括垂直加速和水平加速，乘坐舒适性需要根据调校结果进行调整。不同于现在大多数飞机和火车乘客，我们认为未来的自动驾驶汽车乘客是不会接受较差的舒适度和NVH的。”

电动自动驾驶汽车的另外一个关键问题是如何将电池包的能量损失降至最低。这就需要我们有意识地给所有汽车系统减重。在设计车身架构时，我们需要仔细优化传动系统，包括提高基本机械效率、减少传动损失，同时还应制定可以平衡电机性能、且满足多速传动比需求的系统性解决方案。目前，DSD正在和客户一起为未来的自动驾驶汽车设立标准，并为客户提供架构研发方向上的建议。

Sykes还表示，“如果没有人类驾驶员，另一个传统性能指标——车辆的操控性或响应速度——也就失去了意义。在保证车辆安全性和抓地力的前提下，我们可以通过优化悬挂系统让乘坐更加舒适。为了实现舒适和快速响应两者兼得，传统汽车采用了扭矩矢量控制等传动系统技术，而这些以后都可能会被人工智能算法取代。人工智能算法将使车辆研发阶段的动态包络研究成为历史。”

车速低时，难度不低

低速行驶时，性能令人满意的传动系统能在保证乘坐舒适性的前提下，实现精准操控。在倒车入库或在车流中走走停停时，尤其需要精准地调整离合器，这对于缺少人类直觉的无人驾驶汽车而言，难度很大，“这会影响到传动系统的设计。使用电机电流控制可能会比传统的离合器控制更有效。如果确实如此，那我们必须改良电机设计，使其和变速器设计同步。这里也同样需要系统性的解决方案。”

如果数据太多，人类驾驶员可能会注意力分散，陷入危险。但自动驾驶汽车不同，超宽宽带赋予了车辆极强的数据处理能力，车辆可以一边安全行驶，一边监测来自传感器的海量信息。

Sykes强调，“这使得自动驾驶汽车可以持续不断地处理更多细节性信息，比如动力总成和传动系统的状态。同时，这也为更加智能的设计和轻量化设计提供了机会。制造商可以提前更换即将失效的零部件，不用再为了‘第99个百分位数用户’而过度设计。这种实时监测方案也会改变耐久性标准，连带影响到封装、重量和可服务性。同时，需要指出的是，在自动驾驶汽车领域，我们应该重新定义‘第99个百分位数用户’。”

Sykes认为，除了传统的工程问题以外，业界还需要思考什么样的价位和所有权模式会更适合自动驾驶汽车：是平价的私家车？还是鲁棒性更高、不同标准的营利性租赁模式？和电池包、燃料电池等其它主流车辆系统相比，为了提高使用寿命，业界应制定适合于自动驾驶汽车的动力总成和传动系统策略。

理解负载循环

Sykes还表示，“自动驾驶汽车的各种特性互相影响，这意味着传动系统和底盘工程师未来依然会扮演重要的角色，只是他们的侧重点会改变。在设计传统汽车时，工程师首先考虑的是如何实现最佳性直线能、如何平衡舒适度和操控性。未来，他们将面临全新的挑战和机遇。他们需要用开创性思维去看待自动驾驶汽车的基本负载循环问题。”

“另一方面，除了这些优化挑战以外，在向完全自动驾驶转变的过程中，推进系统的研发工作还是充满机遇的。某个适用于中度自动驾驶级别的解决方案不一定适用于L5。显然，工程师还有很多可以挖掘的空间。”

Sykes的观点得到了DSD美国区总裁JonBrentnall 的认可。Brentnall在考察密歇根州法明顿山的新测试中心时曾表示，“减排是当前业界关注的焦点，这一下子提高了能效的地位，但与此同时，寄生损失等不必要的能量损失依然存在。”

在蒸蒸日上的电动汽车领域，变速器和传动系统一直以来都遭到了忽视，为此，DSD英国测试中心专门研发了尖端测试过程系统，打开了电动汽车变速器和传动系统的应用研究大门，“我们希望为北美市场提供类似的测试服务。”

据悉，DSD的美国测试中心将引进一台负载变速箱效率测试台，未来将逐渐建成三个传动系统试验车间。中心当前正在全面运行的测试台适用于各种类型的变速箱，包括机械增压变速箱等发动机辅助驱动变速箱。该测试台将主要用于测试变速箱的效率，而测试数据也将进一步提高内部变速器效率数学模型的相关性。

该中心计划在今年扩充设备，包括一台用于研发液压阀阀体的液压测试架、一台可以改进润滑油流量分析的斜台测试仪。Brentanall强调，“我们很高兴能为有志于应对未来挑战的工程师提供这个机会。”

测试中心的首个项目，是为一家北美汽车制造商测试研发适用于前轮驱动车辆的并联式混动变速器。

Drawing up a specification for a transmission or driveline may be relatively routine for a conventional vehicle, but for an autonomous vehicle (AV), it's anything but routine, claims automotive consultancy executive Lee Sykes.

“The way we design AVs may require an entirely new approach. We have to address the different requirements of progression through the various SAE Levels of autonomy to understand fully each transition from 1 to 5 and their impact on the functional priorities in order to apply appropriate solutions,” explained Sykes, Commercial Director of driveline engineering consultancy, Drive System Design (DSD), which recently opened a new test facility in Michigan to help vehicle manufacturers and their suppliers increase the efficiency of conventional, electric and hybrid drivelines. DSD specializes in the engineering, development, test and control of future transmission and future driveline systems.

Deciding whether to concentrate on Level 5 or the interim autonomy steps will affect how much vehicle architecture will be carried over and how much purpose designed, he said. But he added that a clearer picture emerges when designing for fully autonomous vehicles where the systems can be fully optimized, without the constraint of accommodating legacy systems.

Transmission and driveline design is steered by a number of key attribute targets defined in the specification, including package, durability, weight, efficiency, NVH, performance, safety comfort and cost. Traditionally, some of these are driver demanded but, if there is no driver, the attribute may have no relevance or, at the very least, the relative weightings of the design priorities will change substantially.

The package demands for an AV transmission will be heavily influenced by cabin space requirements; in the extreme case of a “skateboard” platform, the most suitable solution may involve wheel motors, each with integral reduction gearing. Package size is inevitably related to durability and the required duty cycle.

A summoned-on-demand taxi/rental AV could operate 24/7, clocking up a very high mileage and greatly increasing the duty cycle; on the other hand, eliminating the human driver will avoid mechanical (and passenger verbal!) abuse, or frequent demands for peak performance, so reducing its duty cycle, stated Sykes.

Controlling NVH, refinement and noise have long been a concern for transmission and driveline engineers but a connected AV with an electric powertrain brings additional challenges in an otherwise quiet cabin: “And with no manual driver to blame for any sudden  irregularities in acceleration deceleration, smooth progress will be expected at all times. Refinement expectations will include vertical as well as horizontal accelerations, so ride comfort will need to be tailored accordingly. Although the majority of aircraft and train passengers accept relatively poor levels of comfort and NVH, we believe AV users will not.”

Minimizing energy drain from the battery pack will always be an AV priority, which in turn drives a need for vigilant weight reduction of all the vehicle’s systems. Architecture decisions will need careful optimization of the transmission in terms of basic mechanical efficiency and actuation losses, together with a systems approach to balance electric motor performance with the need for multi-speed ratios. DSD is working with its customers to map out future requirements and advise on architecture direction.

Sykes added: “Another conventional aspect of performance, the handling or responsiveness of the vehicle, is largely irrelevant if no driver is in control; provided the vehicle has safe and adequate road holding, the suspension can be optimized for ride comfort. Driveline technologies, such as torque vectoring, used on manually driven vehicles to combine safety with responsive handling, may well be replaced by AI algorithms which prevent the vehicle from ‘exploring’ the full dynamics’ envelope in the first place.”

Easy going takes effort, too
At lower speeds, satisfactory performance includes the ability to carry out maneuvers in a controlled yet timely manner, without discomfort to the occupants. Shuttling into a parking space or inching forward in stop-start traffic typically require precise clutch modulation, which will be more challenging for a robot driver with no human intuition: “This has implications for powertrain design; it may be more effective to use motor current control than conventional clutch control. If so, it will be important to evolve the motor design concurrently with the transmission – again a systems approach is necessary.”

While a human driver might become perhaps dangerously distracted by too much data, an AV has much greater bandwidth available for data processing, so can monitor far greater input from sensors while still driving safely.

Sykes stressed: “This enables an AV to investigate and process more detailed information continually, such as powertrain and driveline condition. It also permits more intelligent and lighter design, based on pre-emptive replacement of failing parts, rather than over-engineering for the 99^th percentile user. This approach to monitoring in-field life will impact durability specification and have a knock-on effect on package, weight and serviceability. It should also be noted that what constitutes a ‘99^th percentile user’ will need a re-think in the world of AVs.”

Sykes believes that in addition to classic engineering questions, the industry also needs to consider the price/ownership models most relevant to an AV; should it be designed as a low cost consumer product for individual ownership or a more robust commercial product, made to different standards as a revenue generator on a rental basis? An appropriate strategy must be adopted for powertrain and driveline longevity compared to the other major vehicle systems, such as battery packs or fuel cells.

Understand duty cycles
“The interactions between the different attributes of an AV mean that the driveline and chassis engineers will still have important roles, but with different priorities to today,” he said. “Instead of maximizing straight line performance and juggling comfort against handling, there will be new challenges and opportunities. The fundamental duty cycles for an AV will require much future thought too.

In addition to these optimization challenges, understanding how the propulsion systems are developed during the transition phases to full autonomy also brings opportunities. What is appropriate for modest levels of autonomy will not be the best solution for SAE Level 5. Clearly there will be plenty for engineers to get their teeth into.”

His words are echoed in the U.S. by Jon Brentnall, President of DSD Inc. speaking at its new test center in Farmington Hills, Michigan: “The current focus on real-world emissions means the efficiency challenge has suddenly become substantially more critical, yet parasitic and other losses are still draining energy unnecessarily.”

DSD’s UK test center has developed highly advanced processes and systems to ensure that transmission and driveline areas that have not previously received sufficient attention can now be investigated for application in the burgeoning EV world: “It is our intention to build similar test capability tailored to the North American market.”

The U.S. facility will initially house a loaded transmission efficiency test rig and will be developed to include three driveline test cells. The current rig, which is fully operational, is suitable for all transmission types, including engine accessory drives, such as supercharger gearboxes. It will largely be used for transmission efficiency testing and the data produced will also ensure that in-house transmission efficiency math models are well correlated.

Further expansion throughout the year will include a hydraulic test stand for hydraulic valve body development and a tilt rig, which provides enhanced lubrication flow analysis capability. “We are delighted to be offering this opportunity for aspiring engineers looking for their next challenge,” underlined Brentnall.

The facility’s first project is the test and development of a full parallel-hybrid transmission for a front-wheel drive application for a North American vehicle manufacturer.

Author: Stuart Birch
Source: SAE Automotive Engineering Magazine