随着自动驾驶技术研发的不断细化，如何解决“晕车”问题，也成为工程师需要研究的课题之一。高仿真级别的汽车模拟器可以协助身处其中的“驾驶员”体验真车中的各种感受，包括晕车。为了保证自动驾驶汽车的乘客不再遭遇“快停车，我晕的不行了”的困境，如何克服“晕车”，已经变得至关重要。

Ansible Motion是一家位于英国的模拟器专家公司。公司技术联络经理Phil Morse援引密歇根大学的近期研究结果称，在某些特定情况下，比例高达31%的成人均可能在自动驾驶汽车中感到不适。

“其他研究预测中的结果甚至更高，”Morse表示，“英国考文垂大学就曾指出，自动驾驶汽车中的晕车问题‘无法回避’。”

简单来说，一旦车上人员的视线离开道路，就可能会发生晕车，比如读取短信、使用笔记本电脑、观看视频或玩游戏等。但现在的问题是，在自动驾驶汽车的行程中，上述这些行为都将非常常见，就连驾驶员也不会再一直盯着路面。

你有模拟器应激综合征（SAS）吗？

总体来说，影响乘客是否晕车的因素有很多，从设计方面包括车辆的道路颠簸传输频率；噪声、振动和舒适性（NVH）特性；以及具体车型的视野大小等。未来，由于自动驾驶汽车的乘客无需再继续关注车辆操作或周边环境，一些前排乘客甚至还会在高速路段背向车辆行驶方向乘坐，因此晕车问题可能会更加严重。

Morse解释说，“从本质上讲，晕车主要是因为人的眼睛与前庭系统的感知出现差异，也就是说人眼可能已经看到了运动，但前庭系统仍并未感受到这种运动，此时就会出现感知差异，反之亦然。”作为一家汽车模拟器专家公司，Ansible Motion正是在这种背景之下推出了一款DIL汽车模拟器。

Morse表示，公司在DIL模拟器的研发中，投入了大量精力进行虚拟试驾，并最终决定将从机器响应、图像和驾驶员反馈系统入手缓解晕车问题。他说，“目前，汽车厂商已经开始逐步利用模拟器技术，全方位对抗晕车问题。”

为了在驾乘和操控体验之间做出最佳平衡，汽车设计师和工程师做出了巨大的努力。有时，车辆悬架太松或太紧也会导致晕车的发生，特别是在一些体积较大或车身防翻滚性能不佳的车型中。此外，车辆空调性能不佳、座椅结构问题也有可能加重乘客的晕车症状。

与多轴航空模拟器中的宇航员一样，汽车模拟器内的驾驶员也有可能受到模拟器应激综合征（SAS）的困扰。Morse表示，“只要驾驶员接收到的环境反馈出现任何的不一致或延迟，就有可能出现晕车。”考文垂大学发现，高达50%的实验参与者均曾因SAS综合征的困扰而退出测试。”

正因为如此，Ansible Motion公司希望利用自身对SAS综合征的了解与相关经验，解决真实驾驶环境中的晕车问题。比如，公司的工程师可通过调整模拟器设置，专门缓解车内人员的晕车问题，并借此探索车内人员在不同活动下的晕车敏感度及之间的联系。尽管这项测试尚不为大众所知，但对测试结果的分析与理解一定可以给未来的车辆设计提供很多有用信息。

“另辟蹊径”

Morse解释说，随着SAE 3级半自动驾驶汽车及SAE 4级、5级全自动驾驶汽车的出现，驾驶员和乘客之间的身份界限已经开始逐渐模糊，这可能会让晕车成为一个更普遍的问题。考文垂大学的研究人员指出，在非自动驾驶汽车中，大约66%的人曾发生过晕车，且车内娱乐系统也会加重车上人员的晕车问题。

公司的DIL模拟器可提供一个参数可控的独特测试环境，研究人员轻点鼠标即可轻松控制周边环境、天气、车辆性能（物理行为和人体工程元素）及传感器反馈等多个参数，保证测试的可重复性。

设计师可通过尝试不断更换虚拟和实体组件，找到缓解晕车的最佳配置组合。但这种通过改换参数来精确调整驾乘人员体验的方式，仍不够直接。

Morse表示，“即使是将图像延迟控制在一个相对可以接受的范围也需要非常复杂的硬件和软件支持。”目前最复杂的挑战是运动控制。Morse指出，实验室环境并不一定能够复制，或按比例反映真实世界中各种力的相互作用。

为了解决这个问题，Ansible Motion选择了“另辟蹊径”。据了解，AnsibleMotion的DIL系统以一款精心开发人类前庭系统模型为中心，搭配多个“业内独有”的运动控制系统，经过专门设计可模拟大脑对运动、空间及方向等非线性参数的感知。

DIL模拟器系统包含一部可提供6个自由度（6DOF，指刚性物体可在三维空间自由运动的轴数）的分层运动机，并将模拟舱安置在精确控制促动器的顶层上，从而简化了对促动器的要求。

分层运动机的下层提供平面运动，而上层装置则用于产生俯仰、侧倾、前行等运动。正因这种设计，DIL模拟器的重心比一般的六足模拟器（常见于航空领域）更低。

Morse表示，“模拟器的主轴均由单促动器控制。在这种设计下，各种力的控制更加轻松，线性表现更好，且惯性也低的多，非常适合敏感的汽车物理模型。”据了解，Ansible Motion有3家F1客户，该系统最初也是为赛车运动设计的，因此对方向和稳定度的要求非常高，哪怕是进行最细微的调整，也必须向驾驶员提供最准确的反馈。

Corum Technology是一家专攻底盘动态与感知测试的公司。该公司汽车动力学部主管Tim Roebuck对Ansible Motion的系统进行了采样处理。Roebuck指出，用户可随时调整模拟器的物理和反馈参数，“因此，如果我要评估开车或乘车时的舒适程度变化，以及造成这种变化的原因，这种模拟系统比真车测试快很多，几乎可以模拟所有可能的情况。”

Roebuck表示，从最“正常”的汽车响应，到各种更加“极端”的情况，都可以利用这款设备进行车内情况的模拟，因此“我可以轻松感觉出任何体验优化，以及这与车辆行驶状态之间的关系。”

未来，随着自动驾驶系统的不断发展，此类模拟技术与测试手段将在汽车设计中发挥更大价值。

Motion sickness has become a very real issue for engineers developing and testing autonomous vehicle technologies. Automotive simulators can reach such high levels of realism that they may cause their 'drivers' to experience motion sickness similar to that in a real car. Overcoming the issue is vital for ensuring that autonomous vehicle passengers don't suffer the same 'stop the car, I've got to get out' nausea. 

Phil Morse, Technical Liaison Manager of Ansible Motion, a U.K.-based simulator specialist, cites a recent University of Michigan study which concluded that in some situations, up to 31% of adults are likely to experience significant discomfort in an autonomous car.

“Other studies predict even higher percentages," Morse noted. "One, by the University of Coventry [U.K.], refers to motion sickness in automated cars as being ‘the elephant in the room.’”

The problem starts with occupants take their eyes off the road. Causes of car-sickness include reading and texting, laptop computer use, watching videos and gaming—each a plausible scenario for occupants (including the “driver”) during an autonomous car journey.

Do you suffer SAS?

Design factors such as the vehicle’s road disturbance transmission frequency; noise, vibration and harshness (NVH) characteristics and, depending on the vehicle, the levels of outward visibility are all likely to influence the onset and severity of car sickness. Now, add the potential that the occupants are focused neither on the ride nor their vehicle's surroundings. Those sitting in the front seats may, in the future, even be turned rearwards during highway stretches.

“Essentially, it occurs as a result of a perceived mismatch between the eyes and the vestibular system—when motion is seen and not felt, or vice versa,” explained Morse, whose company specializes in Driver-in-the-Loop (DIL) simulator systems for vehicle engineering, including motorsport.

He said that because so much time is spent inside simulators conducting virtual test drives, recent trends in engineering-class simulator technologies have been aimed squarely at mitigating driver discomfort via more responsive machinery, graphics and driver feedback systems. "Today, OEMs are turning to the use of these simulator sickness-mitigation technologies as a means of investigating car sickness," he said.

Great attention is paid by vehicle designers and engineers to achieve optimum ride and handling combinations. But too much or too little suspension compliance—soft, under-damped ride quality particularly in large cars and poorly controlled body roll—also conspire to cause motion sickness. Inadequate HVAC performance and non-optimal seat structure design may further compound the problem.

Drivers in vehicle simulators may suffer Simulator Adaptation Syndrome (SAS) just as their aerospace-industry counterparts do in aircraft multi-axis simulators. “Even very small amounts of latency and/or mismatch between the various environmental feedbacks, e.g. motion, video feeds, etc., can lead to problems," Morse explained. The University of Coventry found that 50% of participants dropped out of simulation tests caused by SAS."

Because of this, Ansible Motion is working to counter real-world motion sickness, using its knowledge of and experience with the symptoms. For example, company engineers can induce motion sickness deliberately by tweaking the simulator’s settings, creating a useful path to explore human sensitivities while people are engaged in different tasks inside a car. It's hardly a popular test regimen, but analyzing and understanding these sensitivities are useful for informing the design of production vehicle.

Taking a different approach

Morse explains that as semi-autonomous (SAE Level 3) and fully autonomous vehicle (Levels 4 and 5) capabilities begin to blur the lines that separate the in-car experiences of drivers and passengers, occurrences of car sickness could become more prevalent. The University of Coventry researchers stated that in non-autonomous vehicles about 66% of all people have experienced motion car sickness, and that the use of in-vehicle entertainment systems can increase its incidence.

DIL simulators offer a unique environment for investigating these effects because they provide a repeatable, controlled environment in which the surroundings, weather, the car itself (physical behavior and ergonomic elements) and the sensory feedback delivered to the driver/occupant, can be altered with a few keystrokes.

By swapping real and virtual components around, designers can efficiently study the combinations that work best to mitigate motion sickness. But manipulating driver/occupant experiences with the required degree of precision is far from straightforward.

“Even cutting down on the graphical latency to an acceptable degree requires highly sophisticated hardware and software," Morse said. The most complex challenge is the motion control. Simply attempting to replicate or scale down the real-world forces doesn’t necessarily work in a laboratory environment, he noted.

To tackle this, Ansible Motion uses what it describes as a radically different approach. It is centered on a carefully developed model of the human vestibular system mated to “industry-unique” motion control systems, designed to stimulate the brain’s perception of movement and spatial orientation, which is inherently non-linear.

The technology incorporates a six-degrees-of-freedom (6DOF; the number of axes that a rigid body can freely move in three-dimensional space) stratiform motion machine. This device simplifies the actuation requirements by placing the cabin on top of layers of precision-controlled actuators.

The first stages provide ground-plane cueing, while upper layers generate the pitch, roll and ride motions. This results in a considerably lower center of gravity than hexapod simulators (used by the aerospace industry) would provide.

“Forces are much easier to manage and primary axes are governed by single actuators, which gives it linear control authority with far less inertia – a perfect fit for connectivity to sensitive vehicle physics models," explained Morse. The system was first developed for motorsport applications (Ansible has three F1 customers) where subtle steering and stability cues are crucial in providing the right feedback to highly experienced drivers.

Tim Roebuck, Head of Vehicle Dynamics at Corum Technology, a chassis dynamics and subjective specialist testing company, has sampled the Ansible Motion system. He noted that the simulator's physics and cueing feedback can be altered on the fly, "so assessing changes in my comfort level as I carried out driving and non-driving tasks could happen at a much faster pace than any testing in a real car. Adjustments in the simulator are almost infinite it seems," Roebuck said.

He was able to experience anything from ‘normal’ vehicle driving responses to ‘extreme’ variations, "so it was easy to describe any improvements in how I felt, and how those related to the vehicle tuning states.” Use of such technology and testing methods will become more valuable as the auto industry increases its development of autonomous-driving systems.

Author: Stuart Birch
Source: SAE Automotive Engineering Magazine