Making E-scooters Safer with Data Annotation

Oct 7, 2021

E-scooters have become a popular and visible feature of many city streets. The micro-mobility market that these machines have helped to create has grown significantly in recent years, and this trend is expected to continue. Scooter-sharing systems in major cities allow users to traverse smaller areas more quickly than walking.

However, this technology faces continuing challenges as a number of municipalities have questioned the safety of e-scooters, particularly when it comes to interactions with pedestrians and other road users. Unless these vehicles can allay the safety fears surrounding them it is likely that more cities will make the choice to ban them.

Computer vision based AI models may be able to meet this challenge by enabling onboard safety and monitoring systems. This blog will look at four safety issues that are being addressed by machine learning developers. We will also look at the specific ways in which data annotation companies, like Keymakr, can further support these efforts.

Keymakr Demo

Targeting safety issues

There are key safety concerns that developers are looking to ameliorate with the help of computer vision. By improving performance in these areas machine learning could help to cement the status of scooter-sharing systems:

  • Pavement riding: Many cities are concerned about the rise of e-scooter users illegally riding their vehicles on pavements. This dangerous practice can lead to collisions with pedestrians and street furniture. E-scooter users are often moving substantially faster than other pavement users, and may be inexperienced in general on the machines.
    AI models can use lane detection to identify what kind of surface a scooter is moving across. This capability can be combined with audible alerts that can tell a rider when they are on the pavement. Through these consistent cues it may be possible to change the behaviour of users and guide them towards safer routes.
  • Problem areas: Another issue for many city authorities is the use of scooters in pedestrianised zones and other problematic/dangerous areas. E-scooters can be an annoyance if they are present in large numbers in these locations.
    Users may be unaware that they have entered a restricted zone, making enforcement a challenge. AI systems can recognize areas of heavy traffic or high pedestrian numbers. Automated systems could even slow scooters remotely as they approach these areas, alerting users and preventing collisions.
  • Safe navigation: A long term aim for AI in the E-scooter market is automated driving and safety responses. Trials are already being undertaken with object detection systems that can quickly identify pedestrians and slow down or divert scooters. Data from E-scooter fleets can also be fed back to authorities. More information and transparency from micro-mobility companies could help to make city governments more amenable to this technology.
  • Parking: The proper parking of shared E-scooters has been a long standing problem for companies and regulators. If parked outside of designated zones E-scooters can act as clutter and an impediment to other pavement users. Computer vision systems can orientate themselves based on visual clues, allowing them to more precisely identify where and how a scooter is parked.
Image annotation | Keymakr

How data annotation providers can help

Keymakr is a data annotation company that provides labeling and verification services to computer vision AI developers. Providers, like Keymakr, can help E-scooter AI innovators to access effective machine learning training data, and offer key advantages:

  • Creating varied data: It can be hard to access image and video data that reflects diverse and chaotic real world environments. Keymakr’s data creation capabilities allow for the capturing of images and video to meet specific needs.
  • Increasing labeled data volumes: Data annotation services can guarantee  scalable data labeling. This allows AI companies to remain flexible and responsive.
  • Ensuring data precision: Keymakr’s in-house annotation team is led by experienced managers who know how to maintain annotation quality levels.

    Contact a team member to book your personalized demo today.
Great! You've successfully subscribed.
Great! Next, complete checkout for full access.
Welcome back! You've successfully signed in.
Success! Your account is fully activated, you now have access to all content.