Announcing Tip-Over Detection

Nov 27 · 4 min read

For the first time, Skip is introducing tip-over detection technology for dockless mobility and a public dataset on scooter tip-overs.

In 2018, we began designing S3, Skip’s first electric fleet scooter. S3 uses unique sensors and onboard vehicle software to benefit both riders and cities. Earlier this month, we began public testing in Washington DC, and we’ve seen promising results so far.

S3, the first electric fleet scooter designed and built by Skip. Source: Greg Stewart staging them incorrectly parked for this photo.

Tip-overs are the most noticeable problem with dockless mobility. There are several contributing factors, such as careless parking, vandalism, and vehicle designs that fall easily due to a high center of gravity or weak kickstands. The result is at best unsightly, and at worst hazardous for sidewalk users, especially those with disabilities.

A recent community complaint flagged tip-overs blocking a bike lane. Source: Twitter user @iolairemcfadden.

In Washington DC, Skip operates with a permit for 720 scooters. During an average 15-day period, we receive 6 community complaints about poor parking and riding, including tip-overs, or 0.0006 daily complaints per scooter. The rate is low because complaints take time and effort to submit, so we often only hear about the most egregious issues. Instead, community members and city staffers frequently find it faster to pick up scooters out of courtesy or frustration.

The onus should not be on community members. So we built tip-over detection to create automated alerts to respond proactively, instead of responding reactively to public complaints. In the process, we have started to analyze more comprehensive data to improve deployment operations, infrastructure investments, user education, and vehicle design.

Tip detection alerts are sent from S3 to Skip’s alert system, allowing us to determine if action is needed and record historical data and performance.

We rely primarily on a 3-axis accelerometer to detect a scooter’s orientation, the same sensor used by smartphones to switch between portrait and landscape mode. When any S3 detects a tip-over, our system records the vehicle ID, GPS location, and timestamp. Over the last 15 days, we detected an average of 0.49 daily alerts per scooter, or 54 daily alerts from 111 scooters.

Tip-over detection also records when a scooter is placed upright. Of all tip-over alerts we received, 11% were placed upright within 5 minutes, 39% within 30 minutes, 53% within an hour, and 74% within 3 hours.

A time lapse view of all tip-over detection alerts from Skip S3 in Washington DC from November 11 to 25. Larger circles indicate a longer time until the scooter returned upright.

We are starting to analyze data by ward, neighborhood, intersection, time of day, and day of week. Patterns are already being used by our operations team to improve fleet parking and deployment and adjust staffing schedules and routes. The data is available for transportation planners, regulators, and city leaders to improve infrastructure, measure compliance, and make data-driven policy decisions.

Histogram of time until upright, bin = 5 minutes. 11% were corrected within 5 minutes, 39% within 30 minutes, 53% within 60 minutes, and 74% within 3 hours.

We can also look at the correlation between high winds and tip-overs, a common question about dockless mobility. S3 was designed with a low center of gravity and a stronger kickstand to resist falling. Recent daily data suggests this design is working. High wind speeds, ranging from 6 to 15 mph, did not correlate with a higher frequency of scooter tip-overs.

Daily tip alerts per scooter do not seem correlated with high wind speeds. Source for wind data.

Clear sidewalks and reducing the burden on communities are crucial to making scooters a fun and effective transportation option that everyone can support. Our next step is to reduce the number of tip-overs and improve our response time — we’ll share more as we learn. We’ve also publicly posted our dataset for anyone who would like to dig in further.

Tip-over detection is the first of many features enabled by S3’s new sensors and software. Technology developed at Skip is bringing a new level of data reporting, accountability, and safety to micromobility.

If you have ideas about improving scooter parking, data sharing and analysis, or if you would like to learn about joining our technical team, we’d love to hear from you!

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade