The adapter is too Sensitive to Accelerations and Decelerations

  • 56
  • Idea
  • Updated 3 years ago
  • Planned
  • (Edited)
Archived and Closed

This conversation is no longer open for comments or replies and is no longer visible to community members.

*Way* too sensitive. Hard Stops and Hard Accels, IMHO, are set too low. All but the slowest actions results in a bleed-el-leep! I have an '07 Civic I do most of my running around in and perhaps because it's so light, even the slowest decelerations result in the Link bleeping at me. My suggestion: instead of a Binary approach to warnings, (either you did accel or decel too fast) use several (3?) levels based on the accelerometer input (i.e. moderately fast accel, extremely fast, etc.)
Photo of Phil H

Phil H

  • 562 Points 500 badge 2x thumb
  • frustrated

Posted 5 years ago

  • 56
Photo of Amy


  • 68,970 Points 50k badge 2x thumb
Official Response
I've changed this topic to an Idea topic as there are a lot of ways that acceleration and braking thresholds could be approached. Some users actually want *more* sensitive thresholds while others want less sensitive thresholds. It's unlikely that there will be one fix that will make everyone happy. So if we're going to address it we'll need to approach this from the angle of developing new functionality.
Photo of Ted

Ted, Former Employee

  • 1,044 Points 1k badge 2x thumb
Official Response
Hi guys!

First and foremost, thanks for using Automatic and contributing on these forums.

Right now we're focused on delivering the most robust possible experience. The current thresholds are rough vehicle-agnostic heuristics-- 80/20 solutions which will help 95% of our users save on gas and wear and tear. I'm sure it's frustrating to encounter this kind of metric, but bear in mind it's just a parameter which we can always tune to perfection.

As we weed out bugs and ship to more cars, we're gaining enough confidence to start looking at more intelligent, adaptive ways to score driver behavior. I'm personally fascinated by this stuff-- I'm working with tons of vehicle data every day, and have a whole slew of ideas about insights to present to our users. I can't say what the final versions will look like, but I'm imagining metrics which abstract away driver behavior and then compare your car to others of the same make/model/year to suggest tire pressure checkups, a personalized vehicle speed/efficiency graph, comparisons of various makes/models' speed/efficiency curves to help you choose your next car, etc.-- all founded on thousands of miles of actual driving instead of theoretical information from the OEMs.

I'm planning to write a blog post or three which will hopefully shed more light on the matter, but in the meantime, please bear with us, and definitely keep sharing your thoughts!

Photo of Amy


  • 68,970 Points 50k badge 2x thumb
Official Response
It's a challenge to support folks who want more and less sensitive thresholds, however, we have some ideas for how to do this really elegantly.
Photo of Tom

Tom, Former Employee

  • 41,706 Points 20k badge 2x thumb
Official Response
Hi everyone. I have a quick update regarding this topic. Our firmware engineers have definitely been working to refine and improve the hard event threshold, and in firmware version (0.9.19), a substantial issue was addressed regarding "bad data" readings. Here's what was happening before the release: speed readings from vehicles weren't happening every second, so the math to determine the rate of change (which directly impacts the hard event threshold) suffered rounding errors. In other words, these imprecise data readings were causing hard alerts to occur more frequently than they should have been.

Having said that, some issues may still be lingering, so be sure to let us know if things have not improved for you.