*5.1. Route Analysis*

To illustrate how the route analysis system described in the work functions, a detailed analysis of one of the routes cycled by one of the users was carried out. It is a 14.5 km route in the city of Corseaux (Switzerland) whose altitude profile is shown in Figure 7. The route was made with a 750 W electric bike with a configuration of six levels of assist, as shown in Figure 2. First, the route was processed using the previously described algorithm to divide it into segments. After this analysis, the system determined that the route was composed of a total of 200 segments. Next, the system calculated the power that the user must provide and the power that the engine must supply to perform the exercise. As detailed in Section 3.3, in this calculation, it is necessary to consider the slope of each segment and the user's ability level in the system. In this case, the user had ability level 4, so the system determined their average speed to be 19 km/h over the course of the route. Based on these data, levels of assist are

their maximum power from 250 W to 750 W.

**5. Results and Discussion**

*5.1. Route Analysis* 

established for each of the segments. The resulting set of assist levels is shown in Figure 8. Finally, these assist levels are sent to the mobile application and the user begins the route. the system determined their average speed to be 19 km/h over the course of the route. Based on these data, levels of assist are established for each of the segments. The resulting set of assist levels is shown in Figure 8. Finally, these assist levels are sent to the mobile application and the user begins the route.

of each segment and the user's ability level in the system. In this case, the user had ability level 4, so

*Sensors* **2018**, *18*, 220 12 of 21

**Table 3.** Set of users for the case of study. **User Sex Age Hours of Activity/Week** 

For the purposes of this case study, all participants had an electric bicycle at their disposal. All bicycles were equipped with the PAS system and were compatible with the *ebikemotion* application. The bicycle of User 1 was road type while the rest of the users' bikes were MTB type or city bikes. The power of the engines of the different bicycles oscillated depending on their model, ranging in

To illustrate how the route analysis system described in the work functions, a detailed analysis of one of the routes cycled by one of the users was carried out. It is a 14.5 km route in the city of Corseaux (Switzerland) whose altitude profile is shown in Figure 7. The route was made with a 750 W electric bike with a configuration of six levels of assist, as shown in Figure 2. First, the route was processed using the previously described algorithm to divide it into segments. After this analysis, the system determined that the route was composed of a total of 200 segments. Next, the system calculated the power that the user must provide and the power that the engine must supply to

User 1 Male 33 13 h User 2 Male 29 10 h User 3 Male 32 9 h User 4 Female 26 1 h User 5 Male 42 10.5 h User 6 Female 31 1 h User 7 Male 28 0 h User 8 Male 30 0 h User 9 Female 27 1 h

**Figure 8.** Levels of assistance calculated for each segment. **Figure 8.** Levels of assistance calculated for each segment.

Once the route is completed, the data collected over the course of the route are sent to the central server for analysis. In Figure 9, we can see the actual power that was provided by the engine (in blue) and the power provided by the user (in green). First, Figure 8 shows how the amount of power is adjusted to the slopes of the route. A greater amount of power has been provided in the segment from 2 to 4 km, due to the inclination of that part of the route. The same happened in the segment from 9 to 10 km and in the final segment of the route, from 12 to 14.5 km. On the contrary, at the segment starting at 10 km and finishing at 12 km, the power provided is lower since the slope of this part of the route is descending. It is also possible to observe how the user's speed is kept as constant as possible on the slopes, although if the selected speed is very high, the power of the engine does not allow the speed to be maintained. With this system, the user's physical effort is not affected by the profile of the route. Regardless of the profile of the route, speed is kept as constant as possible; this helps prevent fatigue and makes exercise healthier. In addition, a heart rate monitor is employed as a safety measure which allows to increase the level of assistance automatically when the threshold of pulsations per minute is exceeded, as indicated in Section 3.1.2. The total watt hours consumed when the route was completed were 240 Wh: 210 Wh were Once the route is completed, the data collected over the course of the route are sent to the central server for analysis. In Figure 9, we can see the actual power that was provided by the engine (in blue) and the power provided by the user (in green). First, Figure 8 shows how the amount of power is adjusted to the slopes of the route. A greater amount of power has been provided in the segment from 2 to 4 km, due to the inclination of that part of the route. The same happened in the segment from 9 to 10 km and in the final segment of the route, from 12 to 14.5 km. On the contrary, at the segment starting at 10 km and finishing at 12 km, the power provided is lower since the slope of this part of the route is descending. It is also possible to observe how the user's speed is kept as constant as possible on the slopes, although if the selected speed is very high, the power of the engine does not allow the speed to be maintained. With this system, the user's physical effort is not affected by the profile of the route. Regardless of the profile of the route, speed is kept as constant as possible; this helps prevent fatigue and makes exercise healthier. In addition, a heart rate monitor is employed as a safety measure which allows to increase the level of assistance automatically when the threshold of pulsations per minute is exceeded, as indicated in Section 3.1.2.

**Figure 9.** Power apportioned to each of the segments of the analyzed route. The power provided by

the user (in blue) and the power provided by the electric engine (in green).

provided by the electric assistance system and 30 Wh were provided by the user. Therefore, the user

was awarded a final score of 30 points upon the completion of the route.

of pulsations per minute is exceeded, as indicated in Section 3.1.2.

was awarded a final score of 30 points upon the completion of the route.

**Figure 8.** Levels of assistance calculated for each segment.

Once the route is completed, the data collected over the course of the route are sent to the central server for analysis. In Figure 9, we can see the actual power that was provided by the engine (in blue) and the power provided by the user (in green). First, Figure 8 shows how the amount of power is adjusted to the slopes of the route. A greater amount of power has been provided in the segment from 2 to 4 km, due to the inclination of that part of the route. The same happened in the segment from 9 to 10 km and in the final segment of the route, from 12 to 14.5 km. On the contrary, at the segment starting at 10 km and finishing at 12 km, the power provided is lower since the slope of this part of the route is descending. It is also possible to observe how the user's speed is kept as constant as possible on the slopes, although if the selected speed is very high, the power of the engine does not allow the speed to be maintained. With this system, the user's physical effort is not affected by the profile of the route. Regardless of the profile of the route, speed is kept as constant as possible; this helps prevent fatigue and makes exercise healthier. In addition, a heart rate monitor is employed as a safety measure which allows to increase the level of assistance automatically when the threshold

The total watt hours consumed when the route was completed were 240 Wh: 210 Wh were

**Figure 9.** Power apportioned to each of the segments of the analyzed route. The power provided by the user (in blue) and the power provided by the electric engine (in green). **Figure 9.** Power apportioned to each of the segments of the analyzed route. The power provided by the user (in blue) and the power provided by the electric engine (in green).

The total watt hours consumed when the route was completed were 240 Wh: 210 Wh were provided by the electric assistance system and 30 Wh were provided by the user. Therefore, the user was awarded a final score of 30 points upon the completion of the route.

*Sensors* **2018**, *18*, 220 14 of 21

### *5.2. Results Overview 5.2. Results Overview*

that they normally spent cycling.

The results obtained at the end of the four-month case study are discussed below to show the progress of the participants. Figure 10 shows the results obtained by the participants and it illustrates the powers each of the users reached to, over the 16 weeks (four months). The results obtained at the end of the four-month case study are discussed below to show the progress of the participants. Figure 10 shows the results obtained by the participants and it illustrates the powers each of the users reached to, over the 16 weeks (four months).

**Figure 10.** Results obtained by the nine case study participants. of fitness. **Figure 10.** Results obtained by the nine case study participants.

As can be seen, most users have exceeded 300 W, which means that they reached ability level 7

amount of exercise per week. After participating in this case study, the three least active users, have managed to do an average of between 4 and 6 h of physical activity a week. The most active users, such as Users 1 and 8, have increased their activity exponentially, surpassing the number of hours

After analyzing the data provided by the heart rate sensor, it was found that the progress of users who exercised regularly throughout the week was less evident than the progress of those who did little exercise. These data are influenced by factors such as the fitness of the user, the duration of the routes and their difficulty. Users who traveled longer routes with higher slopes, made a greater physical effort and therefore their average heart rate had been affected. However, we can look at the data of the two most representative users of the two groups, users who spent a considerable amount of hours exercising each week and those who did little exercise over the week. Figure 11 compares the mean evolution of the heart rate of these two users over the 16 weeks during which the study had been conducted. In the less active group, User 6 did an average of 1 h of exercise each week. In the first weeks, the user's heart rate was high, however it gradually lowered over the next weeks. On the other hand, the progress of User 1 who did 13 h of physical activity weekly, is not as pronounced. This is because active users' heart rate value tends to stabilize once they have reached a stable level

As can be seen, most users have exceeded 300 W, which means that they reached ability level 7 in the system. These users could cycle flat routes with an average speed of 23 km/h, which implies a high level of physical activity. Only three users (6, 7, and 9) were below level 7 in the system, since they did not exceed the 300 Wh. Considering the data analyzed in Table 3, these users do the least amount of exercise per week. After participating in this case study, the three least active users, have managed to do an average of between 4 and 6 h of physical activity a week. The most active users, such as Users 1 and 8, have increased their activity exponentially, surpassing the number of hours that they normally spent cycling.

After analyzing the data provided by the heart rate sensor, it was found that the progress of users who exercised regularly throughout the week was less evident than the progress of those who did little exercise. These data are influenced by factors such as the fitness of the user, the duration of the routes and their difficulty. Users who traveled longer routes with higher slopes, made a greater physical effort and therefore their average heart rate had been affected. However, we can look at the data of the two most representative users of the two groups, users who spent a considerable amount of hours exercising each week and those who did little exercise over the week. Figure 11 compares the mean evolution of the heart rate of these two users over the 16 weeks during which the study had been conducted. In the less active group, User 6 did an average of 1 h of exercise each week. In the first weeks, the user's heart rate was high, however it gradually lowered over the next weeks. On the other hand, the progress of User 1 who did 13 h of physical activity weekly, is not as pronounced. This is because active users' heart rate value tends to stabilize once they have reached a stable level of fitness. *Sensors* **2018**, *18*, 220 15 of 21

**Figure 11.** The evolution of the user's heart rate during exercise, over the course of the case study. **Figure 11.** The evolution of the user's heart rate during exercise, over the course of the case study.

If we look closely at the percentage of routes performed according to the days of the week, we can clearly see that there are two groups. On the one hand, users whose cycling activity is high and constant on the weekends. In Figure 12, it can be read that users in this first group did their activities between 30% and 65% on Saturday and Sunday. While the activities carried out between Monday and Friday do not exceed 15% on average. This kind of users (Users 1, 2, 3, 5 and 8) have improved their average speed significantly over the 16 weeks. If we look closely at the percentage of routes performed according to the days of the week, we can clearly see that there are two groups. On the one hand, users whose cycling activity is high and constant on the weekends. In Figure 12, it can be read that users in this first group did their activities between 30% and 65% on Saturday and Sunday. While the activities carried out between Monday and Friday do not exceed 15% on average. This kind of users (Users 1, 2, 3, 5 and 8) have improved their average speed significantly over the 16 weeks.

**Figure 12.** Distribution of the routes cycled by users from the first group over the week.

the electrical assistance system.

Figure 13 shows the data of the rest of users, the second user group. These users perform a greater amount of activities over the week than during weekends. In this case, on average, 85% of activities were done between Monday and Friday. This is because the users used their bicycles to commute daily in the city. Cyclists such as User 6 progressed in the number of hours they cycled over the week, from 1 h of exercise a week to an average of 5 h per week, thanks to the support offered by

*Sensors* **<sup>2018</sup>**, *<sup>18</sup>*, 220 *Sensors* **2018**, *18*, <sup>220</sup> <sup>15</sup> of 21 and Friday do not exceed 15% on average. This kind of users (Users 1, 2, 3, 5 and 8) have improved

their average speed significantly over the 16 weeks.

**Figure 11.** The evolution of the user's heart rate during exercise, over the course of the case study.

If we look closely at the percentage of routes performed according to the days of the week, we can clearly see that there are two groups. On the one hand, users whose cycling activity is high and constant on the weekends. In Figure 12, it can be read that users in this first group did their activities between 30% and 65% on Saturday and Sunday. While the activities carried out between Monday

**Figure 12.** Distribution of the routes cycled by users from the first group over the week. **Figure 12.** Distribution of the routes cycled by users from the first group over the week.

Figure 13 shows the data of the rest of users, the second user group. These users perform a greater amount of activities over the week than during weekends. In this case, on average, 85% of activities were done between Monday and Friday. This is because the users used their bicycles to commute daily in the city. Cyclists such as User 6 progressed in the number of hours they cycled over the week, from 1 h of exercise a week to an average of 5 h per week, thanks to the support offered by the electrical assistance system. Figure 13 shows the data of the rest of users, the second user group. These users perform a greater amount of activities over the week than during weekends. In this case, on average, 85% of activities were done between Monday and Friday. This is because the users used their bicycles to commute daily in the city. Cyclists such as User 6 progressed in the number of hours they cycled over the week, from 1 h of exercise a week to an average of 5 h per week, thanks to the support offered by the electrical assistance system. *Sensors* **2018**, *18*, 220 16 of 21

**Figure 13.** Distribution of the routes cycled by users from the second group over the week. **Figure 13.** Distribution of the routes cycled by users from the second group over the week.

The web application took record of the days that users checked social statistics on their accounts. The statistics provided by the web application can be seen in Figure 14. In general, all users monitored their performance on the platform. The number of times users accessed these statistics was low in the beginning of the study, however it started increasing with time, especially in the final weeks. Specifically, the activity of Users 1, 2, 3, 4, 8 and 9 increased significantly towards the end of the study and Users 3, 4, 7 and 8 were increasing the number of their cycling activities because they were competing with other users for a higher score. This increase in activity can be seen in Figure 10 in the last weeks of the study. The web application took record of the days that users checked social statistics on their accounts. The statistics provided by the web application can be seen in Figure 14. In general, all users monitored their performance on the platform. The number of times users accessed these statistics was low in the beginning of the study, however it started increasing with time, especially in the final weeks. Specifically, the activity of Users 1, 2, 3, 4, 8 and 9 increased significantly towards the end of the study and Users 3, 4, 7 and 8 were increasing the number of their cycling activities because they were competing with other users for a higher score. This increase in activity can be seen in Figure 10 in the last weeks of the study.

**Figure 14.** Number of days over the duration of the study in which users consulted social statistics.

last weeks of the study.

*5.3. Case Study Limitations*

**6. Conclusions**

important finding of this work.

**Figure 13.** Distribution of the routes cycled by users from the second group over the week.

The web application took record of the days that users checked social statistics on their accounts. The statistics provided by the web application can be seen in Figure 14. In general, all users monitored their performance on the platform. The number of times users accessed these statistics was low in the beginning of the study, however it started increasing with time, especially in the final weeks. Specifically, the activity of Users 1, 2, 3, 4, 8 and 9 increased significantly towards the end of the study and Users 3, 4, 7 and 8 were increasing the number of their cycling activities because they were

**Figure 14.** Number of days over the duration of the study in which users consulted social statistics. **Figure 14.** Number of days over the duration of the study in which users consulted social statistics. *Sensors* **2018**, *18*, 220 17 of 21

Finally, Figure 15 shows a screenshot of the *ebikemotion* web application, where the results of the system can be visualized. On the bottom right-hand side, the rankings of friends who are also users can be viewed. In addition, it is possible to visualize other parameters such as the routes cycled, the calories burned or the scores obtained and the user's ability level. Finally, Figure 15 shows a screenshot of the *ebikemotion* web application, where the results of the system can be visualized. On the bottom right-hand side, the rankings of friends who are also users can be viewed. In addition, it is possible to visualize other parameters such as the routes cycled, the calories burned or the scores obtained and the user's ability level.


**Figure 15.** Screenshot of the visualization of results on the *ebikemotion* web application. The names of the users have been erased for privacy reasons. **Figure 15.** Screenshot of the visualization of results on the *ebikemotion* web application. The names of the users have been erased for privacy reasons.

have an influence on the mode of transport chosen by citizens. It is also important to note that users have not travelled the same routes. Each user travelled different routes with different characteristics. For this reason, it is not possible to directly compare routes with others. The value of the heart rate sensor depends on the physical activities performed by the users. Those users who have performed activities that demand more physical activity, have higher HR values than users with simpler routes.

In this work, a personalized intensity level system for the users of assisted electric bicycles has been designed and implemented. The designed system establishes different assist levels in a personalized way, considering the profile of the route, the power required and the user's ability level. As the user travels new routes, the system awards them with higher scores. The higher the score, the greater the average speed on the routes cycled by the user and the greater the amount of power that the user needs to generate. Thanks to the progressive increase in speed, the user gradually does more physical exercise, improving and increasing their fitness. Therefore, it is possible to replace the gym with the use of the electric bicycle for daily commutes saving economic and time costs. This is an

The innovative component presented in this work is the personalized calculation of exercise for electric bicycle users. Thanks to this system, the user will be able to cycle the routes according to their physical state and ability level. As the user moves up the designed ability levels, the cycling difficulty increases. As demonstrated in Section 5.1, where a route cycled by one of the users has been analyzed, the performance of the designed system is satisfactory. It can segment the route according to its slopes and establish the power that is to be provided by the user, according to its characteristics. The proposed system also evaluates the data collected along the route that had been cycled. This study also demonstrated that the amount of hours the nine case study participants spent on physical activities in a week increased over the four months. This improvement was achieved for both users who were physically fit and those that were not. In general, all users said they were satisfied with

It should be noted that the case study participants were located in four different cities, in three
