- V1:
- Background
- Brief
- Problem
- Goals
- Hero Keyframes Comparison
- Research
- Information Architecture
- Ideation/low-fidelity
- Final Design in Video
- Future Opportunities
- V2:
- Background
- Brief
- Problem
- Goals
- Next-Generation xxx service
- Future Opportunities
- V3:
- Next-Generation xxx service
- 动态或静态工资激励算法
- 实践中, 除了贪婪匹配, 最常用批匹配(batch matching). 平台一般在一定时间间隔内(2-10s)批量进行订单和空车的二部匹配. 包括两个超参数: 匹配间隔; 每个 batch 中允许的最大接客时间/距离(即匹配半径)。
- Global Optimal Solution for Ride-Sharing with EV
- Background
- What:
- Which:
- Who:
- When:
- Where:
- How:
- Future Opportunities
- Possible text from AI
- Draft 1
- Global Optimal Solution for Ride-Sharing with EV
- Background
- Driving Efficiency in EV Ride Sharing
- Battery in the Cloud
- How the connected battery solutions work?
- Why is smart rider-driver matching crucial in the era of electric vehicles?
V1:
Background
产品定位： 专用于网约车的车联网,软硬件一体化解决方案,同时面向To B & To C
Brief
Didi 提高运营效率
Problem
- 续航：（90%以上司机对于续航不满意，期待实际续航达到350km 预计约等于NEDC综合工况400km）期待一天一或两充。
- 充电：充电排队，充电时间规划不合理，周围配套设施不了解，夏天充电慢、冬天充不进电。
- 派单：系统派单不知道车辆的续航 派给司机超过续航的订单
Goals
- 用大数据支撑的Ride-Sharing司机高效派单体验 （续航-充电-派单之间的平衡）
Hero Keyframes Comparison
Before getting into our design process, here's a direct comparison of the redesign: An efficient ride-sharing driver dispatch experience supported by data, balancing the relationship between driving range, charging, and order assignment.
Research
We conducted: Survey + in-persons interviews + online threads
图标结果
Information Architecture
Ideation/low-fidelity
Final Design in Video
Future Opportunities
V2:
Background
产品定位： 专用于网约车的车联网,软硬件一体化解决方案,同时面向To B & To C
Brief
Didi 提高运营效率
Problem
- 续航：（90%以上司机对于续航不满意，期待实际续航达到350km 预计约等于NEDC综合工况400km）期待一天一或两充。
- 充电：充电排队，充电时间规划不合理，周围配套设施不了解，夏天充电慢、冬天充不进电。
- 派单：系统派单不知道车辆的续航 派给司机超过续航的订单
Goals
- 用大数据支撑的Ride-Sharing司机高效派单体验 （续航-充电-派单之间的平衡）
Next-Generation xxx service
- 根据车辆续航的合理派单 （如何展示后台？）
Future Opportunities
V3:
Next-Generation xxx service
In summary：
Feature 1:
- 预约充电 抢单大厅list改变
Feature 2:
- 顺路：抢单大厅list改变
Feature 3:
- 根据车辆续航的合理派单 （如何展示后台？） 侧重Research
动态或静态工资激励算法
- 一些研究只是将工资和补偿集中在供应方面.
- 为了解决供需之间的时空失衡，平台通常有各种形式的时空激励计划。一种常用的时空激励是 Uber 的 Boost 和滴滴的 PanGu: 指定时间段/指定热点区域内的所有行程，成倍提升司机收入。
- Boost 计划在整个城市的不同区域运行，而驾驶员可以获得多少 Boost 取决于他们开车的时间和地点。
- 另一种常用的奖励计划是连续奖励或连续旅行奖励。当司机在指定时间内完成指定热点内的多次行程时，他们将获得连续奖励，金额取决于满足特定要求的行程次数。
- 在短短一周的时间内，乘客激励措施比类似的司机激励措施更有效。从长远来看，例如超过三个月，情况正好相反：司机激励比乘客激励更有效。
实践中, 除了贪婪匹配, 最常用批匹配(batch matching). 平台一般在一定时间间隔内(2-10s)批量进行订单和空车的二部匹配. 包括两个超参数: 匹配间隔; 每个 batch 中允许的最大接客时间/距离(即匹配半径)。
匹配间隔越长, 可调度的车辆和订单越多, 但是等待时间也更长, 导致取消订单; 匹配半径越大, 可调度数量也越多, 但是接客时间更长.
Why is smart rider-driver matching critical for Didi success?
What is so complicated about matching?
The core of the entire order dispatching algorithm revolves around overcoming the uncertainty of future supply and demand, modeling the dynamic spatio-temporal structure, and handling the unpredictability of user behavior. We are increasingly employing deep learning methods to model and predict our spatio-temporal data and user behavior to address these uncertainties. Moreover, compared to traditional ride-hailing, our problem introduces an additional layer of complexity: the issues of electric vehicle batteries, charging durations, and the availability of charging stations. Establishing an accurate simulation system is both a challenge and an opportunity for redefining problems and innovating algorithms in AI for Transportation. Such a system would enable us to develop novel dispatching strategies that optimize for EV-specific factors, inform strategic decisions, and unlock new opportunities for sustainable and intelligent transportation solutions.
How batched matching work?
- 全局最优：The main idea behind order dispatching strategies is to take a big-picture perspective and try to satisfy as many travel demands as possible. The goal is to ensure that every passenger's ride request gets fulfilled quickly and reliably, while also striving to maximize the order acceptance efficiency for each driver. The aim is to minimize the total driving distance and time across the entire system.
- 基于供需预测的分单：Demand-supply prediction-based order distribution: If we predict that a particular area is likely to have higher demand for rides in the near future, then when assigning orders, we encourage more drivers in that area to wait and match with orders originating from the same area.
- 连环派单：After a driver finishes serving their previous order, they are immediately entered into the process of accepting a new order nearby. This effectively compresses the response time for the order and reduces the distance the driver needs to travel to pick up the next passenger.
- When matching orders to EVs, the system takes into account the current battery level and the distance to the pickup location and destination. If the total trip distance exceeds the EV's remaining range, the order is not assigned to that driver.
- Range monitoring(超出电量的驾驶距离不派单): The system continuously tracks the battery level and estimated range of each EV in the fleet. This information is used to ensure that drivers are only assigned orders that they can comfortably complete without running out of charge. When matching orders to EVs, the system takes into account the current battery level and the distance to the pickup location and destination. If the total trip distance exceeds the EV's remaining range, the order is not assigned to that driver.
- Charging station integration （充电站availability实时）: The dispatching system should have access to real-time data on the location and availability of charging stations within the service area. This information is used to intelligently route EVs in need of charging.
- Opportunistic charging assignments（充电站路上顺风单）: If an EV needs to recharge and there's a charging station en route to a potential passenger pickup, the system can assign that order to the EV. This allows the driver to charge their vehicle while still earning fares, optimizing both charging and revenue opportunities.
- Charging time buffer（充电时间blockage）: When an EV is scheduled to charge, the system blocks off sufficient time in the driver's schedule to allow for a full charge. This charging time is treated as unavailable for order matching, ensuring that the EV has enough downtime to recharge properly.
- Dynamic range updates (行驶物理世界天气拥堵反馈在目的地电量上): As factors like traffic conditions, weather, and driving style can impact an EV's actual range, the system should continuously update its range estimates based on real-world data. This helps ensure that range calculations remain accurate and reliable.
- Provide drivers with accurate trip range and estimates of remaining battery on arrival. Mapbox accounts for live vehicle data, driver behavior, current and predicted traffic, and location context like elevation.
Global Optimal Solution for Ride-Sharing with EV
Enhance the ride-sharing driver experience with electric vehicles through a comprehensive matching solution, optimizing everyone's time from trip allocation to charging.
Background
The D1 project is a unique, all-encompassing ride-hailing solution. It seamlessly integrates cutting-edge hardware and software, making it a perfect fit for business and consumer uses alike. D1 stands out by introducing a specialized electric vehicle, specifically designed to cater to the unique demands of the ride-sharing industry.
The D1 ride-hailing vehicle is an innovation for both drivers and passengers. For drivers, it enhances ride-hailing experience by integrating advanced technology into a sleek design that includes real-time operating systems, navigation, communication tools, and a driver-centric interior that prioritizes comfort and control for long shifts. The vehicle is well-equipped with a reclining 'nap mode' seat, massage and cooling functions, as well as ergonomic storage solutions. For passengers, the D1 transforms the ride-hailing experience into a personal lounge on wheels, offering reclining seats, warm ambient lighting, and an accessible control center for a tailored journey.
The D1's interior design revolutionizes the concept of shared mobility, striking a balance between driver needs and passenger comfort. Focused on the driver's ease, safety, and convenience, it carves out an ergonomic workspace that doesn't compromise on functionality. Simultaneously, passengers are treated to a spacious and premium environment. This thoughtful design approach ensures a unique ride-hailing experience where both drivers and passengers enjoy a sense of personal comfort and security, fostering a harmonious journey for all.
➡️The D1 initiative, collaboration between Didi and BYD, aims to revolutionize the ride-sharing industry by introducing a fleet of purpose-built electric vehicles, catering to both B2C and B2B segments. For riders and owner-drivers (B2C), the D1 offers a spacious, comfortable, and safe travel experience, with amenities like adjustable seats, ample legroom, and advanced safety features. For driver-partners and fleet operators (B2B), the D1 provides a cost-effective and efficient solution, boasting an extended driving range, fast charging capabilities, and seamless integration with Didi's ride-hailing platform.
The vehicle's electric powertrain not only reduces operating costs but also contributes to Didi's sustainability goals. The success of the D1's design lies in its ability to address the diverse needs of multiple stakeholders in the ride-sharing ecosystem, offering a comprehensive solution that benefits drivers, passengers, and the environment alike.
What:
The D1 initiative involves the launch of a fleet of electric vehicles dedicated to ride-sharing. These EVs are designed to be eco-friendly, reducing carbon footprints while providing safe, reliable, and efficient urban transportation.
Which:
With a focus on high performance and passenger comfort, the D1 EVs will feature state-of-the-art technology. These vehicles are expected to boast advanced energy efficiency, responsive driving dynamics, and ergonomic design tailored to enhance the ride-sharing experience for both drivers and passengers.
Who:
The project primarily serves ride-sharing operators and users, offering a greener vehicle choice that doesn't compromise on comfort or convenience.
When:
The D1 EV rollout is planned to occur in two phases: an initial launch phase from 2020 to 2025, followed by a subsequent phase of in-depth evaluation and expansion from 2020 to 2030.
Where:
The deployment of the D1 EVs is targeted at urban centers where ride-sharing is prevalent and where the impact of electric vehicles on congestion and air quality can be significant.
How:
The implementation strategy for the D1 initiative will involve a gradual rollout, beginning with pilot tests in select cities to optimize the vehicles' features and performance based on real-world feedback. This phased approach will ensure that the D1 EVs are well-suited to the operational challenges of ride-sharing in dense urban environments.
Future Opportunities
It's insightful to hear about Davinci's strategic pivot and the ambitious goals you have set. Selling 10,000 cars and enhancing brand recognition are significant milestones that will require a strong marketing strategy and a clear brand message.
It's quite an ambitious endeavor to not only enter the ride-hailing market but to also innovate with a specialized vehicle for that purpose. Let's explore this further. Could you share the core idea that sparked the creation of Davinci and led you down this path? What was the moment or situation that ignited this vision?
The decision to create a standardized experience through branded vehicles is a strategic move that addresses core customer concerns of safety and hygiene. It also seems to align well with your goal of selling 10,000 D1 cars this year.
In this journey of transforming your business model, how has your personal story and background influenced the decisions and direction of Davinci?
Building a service on a foundation of robust data and having the largest network of drivers are certainly strong differentiators.
Trust is a key factor in the success of ride-hailing services.
How do you plan to communicate these strengths to your target audience?
What marketing strategies do you believe will be most effective in building that trust and conveying the unique value of Davinci?
It seems there's a challenge in aligning the drivers' preferences with the efficiency goals of your dispatch system.
Effective communication and education could be key in bridging this gap.
Have you considered implementing incentive programs or educational sessions that demonstrate how following the optimized dispatch can lead to increased earnings over time?
What methods of engagement have you found to be most effective with your drivers so far?
Understanding that time is a precious commodity for your drivers, and that the current incentive programs may not be hitting the mark,
it's important to think creatively about solutions that can work within these constraints.
Reflecting on the task-based compensation system, how might you modify or enhance these tasks to make them more appealing to drivers?
Could there be a way to integrate immediate feedback or rewards that align with their goals and the efficiency of your service?
The challenges you're facing are indeed complex, involving both the practical aspects of using EVs and the financial needs of your drivers.
Considering these factors, it might be beneficial to explore incentives that can directly address the downtime associated with charging EVs.
For instance, have you considered compensating drivers for their charging time, or partnering with charging stations to reduce wait times?
Alternatively, could there be a way to schedule drivers' shifts or routes to align with the availability of charging stations, thus minimizing the impact on their earnings?
That's a strategic approach to addressing the EV charging challenge, leveraging data to optimize the use of charging stations and even providing branded stations for convenience.
It's a clear value-add for your drivers and could be a strong selling point for Davinci's service.
In light of these developments, how do you plan to market these advantages to both current and potential drivers to ensure they understand the benefits of joining and staying with Davinci?
Marketing technical advantages to a diverse audience can indeed be challenging.
It's important to convey the benefits in a way that is easily understood and directly relates to the drivers' day-to-day experiences.
Have you considered using more visual or interactive methods to communicate these benefits?
For example, simple infographics or simulations that show potential earnings with and without the use of optimized charging stations might be more impactful.
Additionally, could testimonials or stories from other drivers who have benefited from these programs help in making the message more relatable?
Possible text from AI
Draft 1
Global Optimal Solution for Ride-Sharing with EV
Enhance the ride-sharing driver experience with electric vehicles through a comprehensive matching solution, optimizing everyone's time from trip allocation to charging.
Background
(小字做侧边栏with图片)The D1 initiative, a collaboration between Didi and BYD, aims to revolutionize the ride-sharing industry by introducing a fleet of purpose-built electric vehicles, catering to both B2C and B2B segments. Designing for both B2B and B2C stakeholders, each with their unique needs and expectations, poses a considerable challenge in striking a balance between operational efficiency and user experience. For riders and owner-drivers (B2C), it offers a driver-centric seat with nap mode and massage functions for long shifts, while transforming the passenger space with reclining seats, soothing ambient light, and a control center that lets passengers customize their journey. For driver-partners and fleet operators (B2B), the D1 provides a cost-effective and efficient solution, boasting an extended driving range, fast charging capabilities, and seamless integration with Didi's ride-hailing platform.
In this project, we introduce Didi D1, a complete solution for electric ride-sharing. This platform smoothly integrates intelligent range calculations, optimized batch matching, and charging station management, addressing challenges unique to full-time ride-sharing drivers using electric vehicles (EVs).
Our survey （调研链接） of ride-sharing platform drivers revealed that range anxiety is a major barrier for those considering leasing or buying an EV. The primary worry for these drivers is the system's accuracy in trip planning and dispatching orders in relation to the vehicle's battery range. Drivers often face situations where the system assigns them trips that exceed their current driving range, leading to anxiety and potential service disruptions. Another crucial issue is the availability and waiting time at charging stations. Full-time drivers often work long hours, usually ranging from 8 to 13 hours per day. As a result, time spent in queues at busy charging stations translates to a direct loss of potential earnings. Thus, efficiently locating and using charging stations is of paramount importance to these drivers.
The D1 platform resolves these issues by providing accurate range calculations based on real-time battery levels, driving conditions, and charging station availability. Consequently, full-time ride-sharing drivers can confidently accept trips, secure in the knowledge that the system has factored in their vehicle's range and charging needs. The platform intelligently plans routes that include necessary charging stops, ensuring that drivers can complete their assigned trips without depleting their battery. By optimizing the charging schedule and location selection, the D1 platform minimizes detours and charging time, enabling drivers to boost their productivity and earnings.
Driving Efficiency in EV Ride Sharing
Didi's dispatch system revolutionizes EV ride-sharing by seamlessly integrating vehicle, ride request, and traffic data to accurately predict battery levels for upcoming trips. This eliminates the need for drivers to manually search for charging stations and alleviates their anxiety about having sufficient range to complete ride requests. By continuously monitoring each vehicle's battery level and estimating the remaining range based on driving conditions, route elevation, and historical battery performance data, the system ensures that drivers are better prepared to meet passenger demands without the risk of mid-trip battery depletion, ultimately enhancing the reliability and efficiency of the service for both drivers and passengers.
Gif 1: 热力图 (Gif旁边小字)
We design heat maps that enhance operation efficiency and user experience. These maps show high-demand areas with color gradients, helping drivers identify potential hotspots. This feature allows route optimization, reducing idle time, and increasing passenger pickups. The map updates dynamically with real-time data, providing drivers with the latest demand information.
Gif 2/3:基于车辆电量的派单系统
Our system features intelligent batching and matching that optimizes trip assignments based on the driver's battery level and range. It calculates the driver's remaining mileage based on their current charge and refrains from sending ride requests exceeding this distance. The system also alerts drivers proactively when their battery level is low, indicating that it may affect the range of trip requests they can accept. This user-focused design ensures a smooth experience for drivers and riders, reducing the likelihood of trip cancellations due to insufficient vehicle range.
Battery in the Cloud
Didi has developed a comprehensive, modular portfolio of cloud-based services that enhance battery performance and longevity. This approach improves electric vehicle usage by ensuring batteries operate efficiently throughout their extended lifespan. For drivers, this means reduced downtime and increased earning potential. They can have confidence that their vehicle's battery is managed optimally, lowering the likelihood of unexpected breakdowns or charging problems. Fleet operators can continuously monitor battery health and performance. Through proactive maintenance based on battery data analysis, they can predict when maintenance is required. This allows scheduling during off-peak hours, minimizing vehicle downtime and preventing battery-related issues from affecting service.
How the connected battery solutions work?
Didi's D1 platform uses advanced algorithms to suggest optimal charging locations for rideshare drivers, taking into account charger availability, battery charging curves, and expected local ride demand. The system also calculates the battery's state of charge at the end of a trip when riders are dropped off, factoring in the current battery level, charge depletion rate, ambient temperature, speed, and route gradient. Additionally, it assists drivers in finding the best charging spots and effectively manages the charging network for the entire rideshare community.
Gif 4: Drop-off destination charging finding
The proactive feature optimizes driver experience and fleet management by notifying drivers of low battery levels upon ride completion and recommending the most suitable charging station based on cost, distance, and availability. The AI-driven system calculates these metrics to minimize driver downtime and efficiently distribute drivers across charging stations, optimizing overall fleet performance.
Why is smart rider-driver matching crucial in the era of electric vehicles?
We believe that ultimate efficiency leads to the ultimate user experience. The smart design of batch-matching systems aids the industry in accelerating the transition towards a better user experience in the robotaxi era. As battery technology improves, enabling longer driving ranges and faster charging times, advanced AI algorithms will efficiently match passengers with available vehicles, dynamically optimize vehicle routes, and schedule predictive maintenance. This ensures minimal wait times, reduced empty miles, and maximum fleet utilization. By leveraging vast amounts of data, AI models will continuously learn and adapt, delivering a seamless, on-demand transportation service that is both efficient and user-centric.