Different people drive differently. This could be attributed to a variety of factors such as the routes they take, the road quality, the traffic conditions, the personality of people themselves, and many other reasons. There are some sets of conditions that we encounter very often, such as the commute to office and back, and these comprise a majority of our riding.
A smart vehicle, such as the S340, would allow us to get data from a person’s rides. By looking at many rides together, we can extricate the rider’s typical manner of interacting with the vehicle. The S340 also happens to be a configurable vehicle enabling us to use this analysis to personalise the vehicle to suit the driver’s needs better.
To illustrate the possibiltiies, we recorded some data from different riders to understand how riders throttle and speed. We divided speed into three buckets - low, high, and mid. We then isolated stretches of the ride where riders moved from one bucket to the other in one continuous act of acceleration or deceleration. The proportions between pairs of buckets, and the throttling behaviour. A snippet of our analysis is presented below.
If you just look at how often people move from one bucket to the other, and compare the proportions between different pair of buckets, the differences begin to show up. This information is presented below.
The heatmap of people accelerating from one speed bucket to another is presented below.
Anne, Dick, and Julian have slight similarities in their tendency to accelerate within or to the mid buckets, and occasionally moving from the mid bucket to the high bucket. Julian is the most conservative of the three, with significant low to mid accelerations, Anne completes stretches of acceleration within the mid bracket, while Dick moves to the high speed bracket the most of the three. Timmy and George exhibit completely different behaviours. Timmy seems to move to the mid speed bucket a lot, and stay there, but tends to remain shy of the high speed bucket showing even lesser instances of such an acceleration than Julian, Anne, or Timmy. George, on the other hand, manages to move to the high speed bucket most of the times, and staying there, almost ignoring the mid speed bucket.
The deceleration heat map exhibits similar traits as the acceleration heat map.
Once again we see Anne decelerating within the mid bucket, Dick decelerating from the high to mid, and mid to low buckets, and Julian decelerating from mid to low bucket. George’s style of accelerations to and within high speeds are complemented by his decelereation from and within high speeds. Timmy again shows his tendency to move sharply between the mid and low speed brackets.
In a motor, at a particular RPM there is an upper limit to the torque that can be delivered. In simpler ( although less accurate ) words, at any speed, there is a maximum acceleration the motor is capable of delivering. The throttle settings relate the amount of throttle being applied by the rider to the proportion of the maximum torque the motor should deliver. In simpler terms, the throttle controls the amount of acceleration you experience by specifying it as a proportion of the maximum acceleration the motor could have achieved at that speed. This is called a throttle map.
The rider usually gets to experience only one throttle map as dictated by the parameters set by the manufacturer of the vehicle. However, a smart vehicle allows the possibility of reconfiguring the throttle response by modifying those parameters. Furthermore, a smart vehicle could have multiple throttle maps which the user could switch between or the scooter could intelligently switch between dynamically. With that intent, and with the understanding of the acceleration behaviour of our riders, let us look at how they use the throttle. Deciding the actual settings and switching behaviour is a complex problem and would need to consider the safety and the comfort of the rider apart from other factors; this is a simplified picture to illustrate the possibilities.
The plots below chart out the cumulative usage of the throttle during the acceleration and deceleration stretches. The bucket pairs which didn’t occur often enough were filtered out. As an example, in the box labelled acceleration, the lines for Anne and George at x = 70% indicate that they have the throttle rotated by less than 70% for about y = 80% of the time. Note how the lines itself follow different trajectories to reach that point. Since this is cumulative usage, you can see the lines approaching 100% on the Y axis. The corresponding point on the X axis is the highest ever throttle in that box. In the same box, Timmy reaches 100% on the Y axis at a little over 90% which implies that Timmy never throttles the full 100% when accelerating.
Right away, the diversity in the lines indicates that there is potential to customise the throttle. We have already observed that different riders accelerate and decelerate differently and this is reflected in their throttling too. A curve which looks well spread out across the span of the plot indicates a person who uses the full spectrum of the throttle. Such a person probably finds the resolution of the throttle response to be acceptable and would be okay with the current throttle map that they are using. A very skewed curve would indicate very narrow usage of the throttle. Such a person would appreciate greater resolution of the throttle response in the band that they occupy at the cost of lower resolution of the throttle response in the bands that they don’t occupy.
Timmy would be a great candidate for a customised throttle map. Given how he tends to stay in the low throttle bands for a large majority of the time, he might like finer control in that region at the expense of lesser resolution in the higher regions. We could lower the proportion of maximum torque delivered which would Timmy to use the entire span of the throttle and offer him better control. Anne’s usage of the throttle tapers off after 70% only to suddenly rise at the 90% mark. A customisation for Anne might include higher proportions of torque between 70% and 100% to spread the spike at 90% to the nearby areas. Dick tends to be more aggresive than Julian about halfway through the span of the throttle after which Julian becomes more aggressive. At least one of them, if not both, could use slightly altered throttle maps on either side of this intersection point.
Now imagine being able to do this analsis and calculation in greater granularity. By allowing different user profiles on the same scooter, each rider has the ability to have a throttle map they are most comfortable with. By looking at these numbers within particular speed bands and assigning new throttle maps within each speed band, we could simulate the sort of drive the rider experiences on a typical petrol scooter, which has a CVT mechanism. You could be more demanding and have different maps for certain times of day, days of week, or even for locations. It’s possible.