• Improve HMI for ride-hailing efficiency in D1

Improve HMI for ride-hailing efficiency in D1

D1 and main challenges caused by a mixed business model

D1 is the first purpose-built ride-hailing EV in China that caters to both B2B and B2C markets. It targets at high-end experience sensitive passengers, with a special layout that maximizing back seat space, lugage space, and a side sliding two-direction electric door. Full time drivers can rent a D1 from the fleet company, and use the car for full day operation, while also privately use it as a personal car. This complexity in business model had placed special challenges for its design, to serve for both fleet compnies and drivers, and with needs from fleet management and customer daily use.

The problem started to arise while it hits 10k orders and went on real world operation in Shenzhen. The team investigated into reasons that prevent drivers from (continuing) rent D1, and it presents complains over lower daily income that indicate inefficiency in daily operations. Drivers also expressed concerns on privacy of exposing operational details(such as income, history,etc) while used in a personal scenario. To solve this, Didi incoperated cloud computing on management over batteries and chargers(called XiaoJu Power), and put up the goal for improving ride-hailing operational efficiency as the next milestone. I was inovlved in the early phase as a UX designer, collaborated with the team made of car designers, product managers and modeling engineers, worked on the improvements on HMI, with the focus to foresee future use case with implemented smart argorithm to improve operational efficiency and add the competivness of this car.

Finding the right problem: what causes the inefficiency

I received the research presentation made by product managers (with original data unavailable), and noticed how they made conclusion on needs divided to three stakeholders: the driver, the passengers, and the car provider (CP) companies. The drivers’ quotes has caught up my attention. Some complains about the poor working environment “have to work over 15hrs on winter and the coldness hurts my feet”, “slept over inside the car to work extra hours and received passengers’ report for bad smell”. Others complains about the charging issues “have to wait in lines to charge” “hesitate and refused to take long-range orders because of charging wories”. This has indicated a poor time management in driver’s operating hours after transfering to EVs, as the charging related planning interfer with route planning, adds up with the challenges from the close to saturate market, D1 drivers might compensate with extra working hours. Charging anxiety creates a significant disadvantage towards driver’s decision in renting D1.

The research over screen functionalities also provide intersting data. 64% drivers support central control through screens, although 43% expressed acceptance for not having one. The in-car screen was used mainly for car settings (33%), media(33%) and navigation (26%), with the mobile phones was used mainly for information (29%), navigation (29%), and social apps (27%) in comparison. The data impressed me with how in-car screen was weak in functionalities thus drivers relies highly on the phone, allowing me to notice the pattern of charging the phone often during daily operations, with car infotainment systems and AC turned on. Combined with another data showing driver seldomly plays media when having a passengers inside, it is fair to conclude that media and entertainment has very less impact in a working scenario, meaning that the previous design emphsis on providing rich consuming contents and apps has gone wrong.

Combined analysis with other data and calculate its estimation, I pictured a day of a representative full-time driver. They are aged 30-50, majorly males, work 6days per week, spend 8-12hrs with the car, with 2h relaxing time(based on break 20mins every 4hr to prevent drowsy driving, according to Didi drivers’ manual), 2hr charging time (based on the real driving range if 350km, fast charging 20%-80% with 30mins, city speed 30-40km/h, thus charged 2-3 times). 80%in-car time are under work scenario. The money(substract the cost of energy) they make per day is similar to gasoline car drivers if break time and wait time overlaps. Although real world conditions would be significantly different from the estimation, but it indicates the possbility of improving efficiency through a better arrangement.

Creating solution for efficiencies

I want to tackle the problem from two aspects: the real efficiency and the percived efficiency. Real efficiency could mean that the dispatch system better distributed orders so passengers and drivers expereince less idle time across the system, this gives drivers a better arrangement of time “real working”. Real efficiency could also mean that the system better distributed chargers so drivers have a better arrangement over charging. Percived efficiency refers to how the driver feels on the efficiency improvements, thus they consider D1 more competitive when rent a car. It can further break down into the efficiency of commuinication, efficiency of navigation, and the efficiency of use.

Thus, I brought up the systematic method to talckal the problem, combined with other solutions

Making decisions for ride-hailing specific needs

The initial idea starts with a drawing on the paper. When discussing with the other UX desginer, we aggree it would be a good idea to make charging as a seperate tab, thus it is always accessible to drivers, when they have the needs. we also aggree to present price as the primary focus, with the smart suggested order always on top.

The challeges rises when we try to figure out the layout, as it indicates the system logic. From the old design, we observe that it used a hub-and-spoke design model (homepage acts as a central point from which users can navigate to other features). This model is usaly used when the homepage is essential and centralized, in order to organize a large set of features. This design is suitble with the general use, when users spent time in cars and may have needs to access other apps, but this sceanrio is not the major use case for drivers. I proposed we create a seperate home page and use a mixed model of flat navigation and hub-and-spoke to cater for both needs. This idea defines the our slection of functions on the navbar. The app menu is the homepage for apps, and the logo suggest the “homepage” of operation related informations that makeup for the previous miss. The new ‘homepage” will be a flat navigation, with map always displayed as the primary information inquirey entry. This design will also have the benefit for privacy. When all operation related info was kept under the drivers’ tab, it can be easily hide when accessing other features. This also caters the need of private use.

The navbar has also went through intensive dicussions. after we decided on car settings, drivers, apps and charging,. The ariconditioning had been through adjustments to wether hide the AC control from the navbar. The final decision was made when we consider the ride hailing use case, as passengers have diversed needs towards in-car envrionment, expecially for this ride-hialing specific vehicle. For the sequence, we examine its ergonomics based on the given screen size and location, and decided the general upper area will be easier to reach compared to the lower. The charging tab is an function that is frequent in using, but contains less information. it is placed on the bottom corner, leaving enough space between other icons to suggest its specifically, with a location that is relatively harder to reach, but form good muscle memory without looking at it.

The most challenging moment is the decision-making in the information display of the card: which will be the container of smart order assignment, and smart charging suggestions. Taken order assignment as an example, the information related to location are current location, pickup location, dropoff location, while each location acompanied with its time to arrive, distance from the car, and remaining batteires. Compared to the draft I made, the final version prioritize the price and battery consupmtion, with time, address are grouped as one, and clear distinction between now, the order start, and the order finish.

Another intensive discussion happens on a more systematic level, about alternative and unaccepted orders. Our observations showed that drivers will not take the order if they are percieved less “valued”, and this might impact the overall efficiency of the batch. My point of view is we should always leave an alternative option, because this will improve the error tolerance, especially in unexpected real world situations. The final design we adopt is to have a timing bar attatched to the most recomended order, and it will disappear if not accept in a given time. Drivers manually chose from small cards, and The option will be provided to other drivers in the same batch as alternative options (appears as small cards below). This situation refreshed when the next batch matching begins, thus it won’t last for over 2mins.

the map features cause disussiono on the systematic structure as well, that is whether the left panel should always be displayed. The discussion was broughtup as I found out direction information plus left panel ocupies a large area of the screen if they appeared simultaneousy, thus making the direction information less visible, if considered a driving scenario. The direction info cannot appear inside the left panel, as the needs for navigation happens after taking an order, thus communicating with customer could happen during driving. The final decision we agree on was to cut off the feature to select orders while navigation, for safety concerns conflicts.n and intergrate passenger communications inside the direction infor. this will prevent the case that drivers go to other tabs with the left panel, when the direction window is on. They will need to stop the navigation in order to access the left panel.

other features are designed specifically for ride-hailign needs. For example, drivers can complete task to earn credits, view order history and dayliy earnings. These features assembled the core usecases on the mobile app, but not been touched on the old design. I proposed a switch machanism, to allow contents shift between earning information, and order recomendations through “go online”. This also helped in simplified the interactions.

Design impact and deficincies

This design was internal reviewed and tested in a very small range based on number of clicks needed to complete operation related tasks, showed improvement in reduce 1-4 clicks. The feature that receives most appreciations is the heat map on the driver’s page, because it is a significant simplifications of the old mechanism opening the heatmap, that drivers often complained about hard to find. The smart recomendation is also appreciated as it saves decision making energies, but wheter the information displayed is effective have room for a real world application. The limitation of time and conceptual nature also cause its performance in accesibility, major information display passed the contrast test, but not for some sub and minor descriptions. But overall, the feedback is positive from higher managers, especially the aethetic style that presents a robotic and futuristic tech style, which proven to be attrative to the user group.

This design was internally reviewed and tested within a very small scope, using a ipad in a cabin prototype. It demonstrated improvements by reducing 1 to 4 clicks needed to complete general operation-related tasks.

The heat map feature received the most appreciation. The new heat map can be viewed in center HMI — compared to that it was exclusively accessible on phone app, and the feature is easy to find — compared to drivers have been complaining it was hard to find or that they didn't even know it existed because it was buried in an unrelated menu.

The smart recommendation is also well-regarded because it saves decision-making energy. However, to evaluate its effectiveness in communicating information, it still needs review from real driving application, rather than the simulated stationary setting we had when testing.

Time limitations and the conceptual nature of this project also impacted evaluation in accessibility. Major information displays passed the contrast test, but some sub and minor descriptions did not. This suggested a room for improvement in future iterations, to meet the standards for in-car displays.

Overall, the design received positive feedback from higher managers. They liked the aesthetic style. The robotic and futuristic tech look has proven to be attractive to male drivers.

The design is

Although the effectiveness of the information still leaves room for real-world evaluation.

Drivers complained heatmap was hard to find on the phone app,

Compared to the old method for opening the heat map — a function that drivers complained was hard to find,

After an internal review and testing within a very small scope, this design demonstrated improvements by reducing the number of clicks needed to complete operation-related tasks by 1 to 4 clicks. The feature that received the most appreciation is the heat map on the driver's page; it's a significant simplification of the old method for opening the heat map—a function that drivers often complained was hard to find. The smart recommendation is also well-regarded because it saves decision-making energy. However, whether the information displayed is effective still leaves room for real-world application and evaluation.

Time limitations and the conceptual nature of the design also impacted its performance in accessibility. Major information displays passed the contrast test, but some sub and minor descriptions did not. Overall, the feedback from higher managers is positive, especially regarding the aesthetic style. Presenting a robotic and futuristic tech look, it has proven to be attractive to the user group.

this design underwent a small-scale internal review and testing, focusing on the number of clicks needed to complete tasks. The result? A reduction of 1-4 clicks! The heat map on the driver’s page garnered the most praise, significantly simplifying the previously hard-to-find mechanism. Drivers, often frustrated with the old setup, found this new feature much easier to navigate.

The smart recommendation system also received applause for saving decision-making energy. However, its real-world effectiveness remains to be fully tested.

Accessibility had mixed results due to time constraints and the design's conceptual nature. While major information displays passed the contrast test, some secondary and minor descriptions did not. Nonetheless, higher managers responded positively, particularly enjoying the design's robotic and futuristic aesthetic, which appealed to the user group.