In Part 1 of this series, I discussed how poor input is a key source for user frustration. When the input is garbage, the output is garbage. While this holds true for any form of input, audio seems to suffer more than touch in modern cars. Fortunately, we talked about many ways to solve this that are pretty easy – user higher quality microphones, placing them appropriately, and using software to enhance the quality of the audio signal. Once we’ve done this, we have the foundation we need to deliver a satisfactory driver experience. But we can’t stop there. What are some of the other pitfalls that can get in the way?
Pitfall 2: Giving the users too much
If we overload the driver’s attention, we’re asking for trouble. This is something that happens too often in cars on the road today. Systems are often designed for use in optimal conditions, not accounting for a driver’s already high cognitive load from the very act of driving, conversation with passengers, or pondering what to pick up from the store after work. When we give drivers too much information in a single prompt or on the screen, it may require more attention to process than a driver can safely give. In a world full of distractions, we can’t afford to be another one where it isn’t absolutely necessary.
Imagine if I was providing you directions to my house in Concord from my office in Cambridge, MA.
…Turn right on Massachusetts Avenue and then follow it until you reach Harvard Square. When you get there, the road will split and you’ll hang left but stay in the right lane so you can turn right immediately. Follow that road….
Maybe this seems a bit contrived, as most in-car navigation systems aren’t giving that many turns in a single shot. In fact, it does seem a little crazy. And yet, we occasionally see this happen in other parts of the HMI. We tend to overload the user in the car. The problem happens when we provide more than a couple of steps or pieces of information the driver needs to remember, and the driver tunes it out or forgets some of it.
The worst: Help prompts. Some systems may provide information on every domain supported, with a web address for a specific point of interest (POI) and a phone number to call for additional support. For example, in one car I recently evaluated, the help prompt was just over a minute long and I had to listen to it three times to get the piece of information I needed. And because I work in the auto HMI space, I already have it easier than the typical driver who is experiencing these types of systems for the first time.
But this is fixable. We can make the systems conversational by incorporating Natural Language Understanding (NLU). Think about how we get help from someone else. It’s a two-way dialog that requires a lower cognitive load and feels more natural. This works best when help is domain-specific instead of all encompassing. NLU plays an important role because we often ask for help when we’re unsure or confused. NLU allows us to work with the wide range of statements or requests a driver may make when trying to use the system before they’ve learned the best way to phrase something. It makes it so the exact phrasing doesn’t matter.
How to avoid driver information overload when using automotive human machine interfaces
Visually, we risk falling into the same trap of HMI information overload. If we ask a driver what information they want to know at any given time, we’ll end up with a long list that includes everything from phone battery, to a map, to the album art for the current song. Even if each of these elements is highly visible, the driver’s ability to glance at the screen is compromised by too much information.
The solution comes in the form of determining what drivers actually need to know at any given time, and prioritizing that information as the most important. Typically, information not relevant to driving or the current task should be hidden.
Providing the right amount of information for the user is a balancing act between too much and not enough information
On the other hand, overcompensating and presenting too little information is another risk. If we go too far, the driver may have to use the system and dig around for the information he or she needs, resulting in even more distraction and a worse experience.
It’s a balancing act, and one that should be done carefully and informed through research. This is a huge part of what we’ve been focused on at Nuance. We’re investigating what information users consider the most important in every domain in the car, and learning more about how they prioritize this information. Through these investigations, we’re moving toward empirically-driven guidelines that can guide which information we choose to provide to a driver, and which information we hide.
The right amount of information
On the surface, trying to limit how much we distract the driver seems obvious, much like the first automotive HMI pitfall. Yet, often, we fail to account for all the other events occurring for the driver. Our systems should be viewed as conversational systems. The litmus test for almost any in-car interaction should be if we could have the same interaction with another person and arrive at the perfect outcome. If the driver is not able to process the information when provided by a passenger, an infotainment system certainly shouldn’t provide it.
In my next post, I’ll talk about the third pitfall: Mismatched prompts and dialogs.