Original Source Here
The Future of e-Scooter Maintenance and Rider Risk Assessment is in Inertial Sensors
In recent years, e-scooters have become an integral part of the urban landscape and its habitant’s daily lives. The micro-mobility revolution happens too fast for regulation to keep up and poses an opportunity for technological innovation. Municipalities all over the world are limiting the permitted area for rental e-scooter operation, limiting the maximum riding speed, and managing parking space by squeezing whole fleets into magical rectangles.
Road casualties due to collisions between e-scooters, pedestrians, and cars are increasing rapidly worldwide. This raises hard regulatory and technological questions: how can we minimize risk to the rider, the surrounding, and the e-scooter? And when will there be an affordable insurance policy that covers micro-mobility-related injury and damage?
In a series of posts, we’ll elaborate on rider activity detection and e-scooter maintenance monitoring. We’ll explore the use of a widely available sensor that exists on any smartphone and in most e-scooter communication modules (for example check out the fantastic Comodule company from Tallinn). That sensor is of course the inertial measurement unit (IMU).
What is an IMU and how can it be used for rider activity detection and e-scooter maintenance monitoring?
The IMU consists of an accelerometer and a gyroscope. The former measures acceleration and the latter measures angular velocity. The two sensors provide information at a high rate that ranges between 100 Hz and 500 Hz, depending on the device. This means that riding at a speed of 10 m/s creates at least one sample every 10 cm of road. Each measurement contains 6 values: 3 from the accelerometer and 3 from the gyroscope, i.e., one for each of the three axes.
ALMA Technologies, an Israeli company, has recently demonstrated accurate positioning using an IMU sensor for the mobility and micro-mobility sectors. Accurate positioning is achieved by sensing the motion of the car or e-scooter and as a result, the IMU sensor may serve another purpose. ALMA’s products collect and analyze position and motion information to determine rider activity and e-scooter maintenance status. Some examples include low tire pressure detection, riding tandem detection, riding surface recognition, and dropped and deserted electric scooter detection (you can find more details here).
Recently, ALMA has initiated collaboration with Connected Insurance, an Israeli insurtech company that develops data-driven insurance solutions. This collaboration aims to bring smart insurance to the micro-mobility market by utilizing the widely available low-cost IMU sensors for rider activity detection and e-scooter maintenance monitoring.
Leveraging the positioning information for the driver behavior and e-scooter maintenance
Navigation/positioning systems have become an integral part of our daily life. We rely on them to make our lives simpler, safer, and more convenient. However, existing systems cannot operate without proper satellite reception (GPS). ALMA creates positioning solutions that reduce our dependency on satellite navigation, also for the micro-mobility sector. If the fleet manager knows what is happening to every e-scooter in its fleet, anytime, anywhere, he can manage risky behavior for the rider and the e-scooter. Some examples include Tire low-pressure detection, riding tandem recognition, riding surface recognition, and dropped and deserted electric scooter detection (you may refer to the entire list here).
Use Case — detection and classification of turns
Rider activity detection is made possible by extracting motion features from IMU and position data. A basic example of this is turn detection and classification.
In this example, we recorded a typical e-scooter ride and plotted its trajectory in blue (see figure below). To record the IMU and position data, an iPhone was fixed to the handlebar using a regular mount. The speed limit in this scenario was set to 8 m/s.
Using ALMA algorithms, we produced the following plot and analysis. The next figure shows the trajectory of a 3 km ride in blue with colored markers scattered along the route. The markers show positions where turns of at least 60 degrees were detected. The color of each marker indicates the speed at which the turn was taken.
Fast and risky turns may be detected and compared with the typical turn speed in the area of interest. This approach allows fleet managers to learn how their e-scooters are used by continuously collecting data. Moreover, in case of an accident, it becomes possible to analyze rider activity before and during the incident and compare those activities to other drivers in the area. From the maintenance perspective, one can determine the e-scooter energy consumption profile of a rider which may impact ride pricing and e-scooter maintenance scheduling.
The fascinating era of micro-mobility, and e-scooters in particular, has just begun! There are many open questions regarding rider behavior where no one really knows how to manage the e-scooters maintenance efficiently and the rider’s risk- from the insurance perspective. In this post, we discussed some problems. Just imagine what can we do with many features and powerful machine learning!
About the authors
Entrepreneur. AI and navigation expert; Ex-Qualcomm. Barak holds M.Sc. and B.Sc. in Engineering and B.A. in Economics from the Technion. Winner of Gemunder prize. Barak is a Ph.D. candidate (final phase) in the fields of AI and Sensor Fusion. Author of several papers and patents.
Maxim Freydin, Ph.D.
Experienced researcher with vast knowledge in Signal Processing, Applied Mathematics, Structural Mechanics, and Machine Learning; Ex-Elbit. Maxim graduated with a Ph.D. from the Technion and a postdoctoral fellowship from Duke University. Author of several papers and patents.
Trending AI/ML Article Identified & Digested via Granola by Ramsey Elbasheer; a Machine-Driven RSS Bot