Who says the auxiliary driving of micro-cars is difficult to stand out?

Author: Wang Xuan

The Wuling MiNi EV lowers the threshold of automobiles, and the Baojun KiWi EV lowers the threshold of assisted driving.

In Liuzhou, Guangxi, the land has become “magical” due to the presence of Wuling Motors. Fifty percent of the motor vehicles on the road here are micro electric vehicles of Wuling’s brands, such as Baojun KiWi EV, Hongguang MiNi EV, Baojun E100, Baojun E200, and many more. The municipal government has special support policies for these vehicles, which include exclusive parking spaces for micro cars and two-hour free parking benefits.

After the popularity of the Hongguang MiNi EV, many people are reconsidering the relationship between micro electric vehicles and urban transportation scenarios, and there are bound to be many people who are holding their money. Perhaps you may feel that driving such a car on the street is a bit “tacky”; or perhaps you stubbornly believe that “range anxiety” is the biggest obstacle to purchasing a micro electric vehicle.

If that’s the case, then I would recommend that you visit Liuzhou. The incredibly mature market for micro electric vehicles there will definitely infect you and dispel all your concerns. You will also find that when micro electric vehicles are gradually accepted by the market, their next problem to solve is intelligence.

It is not easy to develop an assisted driving system for micro electric vehicles.

The positioning of models without missing scenarios

There are two ways to make an assisted driving system outstanding: 1. Better experience, 2. More functional matches for more use cases. The first path is somewhat difficult because, for consumers, advantages that cannot be reflected on the balance sheet are equivalent to no advantages.

Therefore, many automakers choose to focus on the second path, and the “Highway Navigation Assistance” function has emerged. Compared with urban navigation assistance, it is easier to technically implement, and has more scenarios and selling points relative to basic L2.

For the Baojun KiWi EV, it is obviously not suitable for the second path. As He XPeng said at the S4 ultra-fast charging pile release conference last week, when the selling price of a car is higher and the size is larger, users will give it more missions. With a selling price of less than 100,000 yuan and a maximum speed of only 116km/h, the use case of Baojun KiWi EV is relatively simple.The “Lingxi Smart Driving System” of Baojun is a urban travel solution with the core advantages of “parking scenario” and “congestion scenario”, and it is very clear that Baojun has thought it through.

The smart driving chip of Baojun KiWi EV is from Texas Instruments, and the main perception hardware is the binocular camera provided by DJI, which uses DJI’s visual algorithm.

In terms of parking, Baojun KiWi EV has three advantages:

  • The vehicle does not need to move, and the strong visual system can recognize parking spaces while the vehicle is stationary.

  • “Visible and Stoppable”, narrow parking spaces (including horizontal, vertical, and diagonal parking spaces) 15cm beyond the rearview mirror can be parked safely and steadily, and it can accurately identify and plan the parking path according to the parking space and the actual situation nearby, even facing a corner parking space.

  • “AI Smart Exit Parking”, the parking scene is not troubled by the shape of the parking space and the state of the vehicle, and intelligent exit parking can be achieved in any way, such as intelligent recognition of the vehicle head entering the parking space and exiting with the vehicle tail.

However, for Baojun KiWi EV with a length of 2894mm and a width of 1655mm, parking may not be the most critical pain point for users, after all, we have only seen users complaining about the difficulty of parking full-size SUVs in the city. It cannot be denied that the parking system of Baojun KiWi EV is indeed very powerful, but it has been hindered by the size of the car model, which once again proves that the intelligent driving of micro cars is difficult to shine.

In comparison, the “Lingxi Smart Driving System” performs the best in urban scenarios.

How to make urban assisted driving more comfortable?

In the premise that the city navigation assisted driving has not landed for a long time, how to make assisted driving more comfortable in the city? Baojun’s conclusion is to create a better human-machine driving system.

Human-machine driving can be divided into three levels: information perception level, planning decision-making level, and execution control level.## Information Perception Layer
In urban road conditions, the lane is easily lost by the advanced driver assistance system, for instance, when approaching an intersection. Without a lane, the driver can take over the vehicle’s direction. Once the vehicle detects the driver’s taking-over, it will temporarily exit the control of the direction. When entering the next road segment with a clear lane, the advanced driving assistance system of the Baojun Kiwi EV can quickly detect and automatically regain control from the driver.

(Starting at the stoplight, manually selecting the lane in the middle of the road and smoothly transferring control to the advanced driver assistance system.)

When the driver does not intervene, the vehicle will automatically follow the preceding car’s trajectory. In case of chaotic lane conditions, the Baojun Kiwi EV can identify and analyze its surroundings through visual algorithm to find a more feasible and reasonable driving route and maintain its intelligent driving assistance without exiting. Additionally, the vehicle can automatically change lanes and merge into the main lane through the lane trace on the roadside.

Planning and Decision-making Layer

The planning and decision-making layer is more evident for vehicles equipped with the urban navigation assistance feature. Vehicles can proactively select faster driving lanes in urban road conditions through human-machine co-driving strategies that consider the “optimal route.”

Currently, the decision to select the optimal route is entirely in the hands of the driver as the urban navigation assistance feature has yet to be deployed. The Baojun Kiwi EV stands out from other driving assistance products as it does not excessively intervene when the driver selects the optimal route.

Firstly, the vehicle has the function of automatically changing lanes when the indicator is activated. Meanwhile, the driver can easily adjust the lane manually. When the driver shows a clear intention to change lanes (that is, when the turning force on the steering wheel exceeds 2.0 Nm and lasts for 50ms), the vehicle’s lane change function will return to the driver. Then, the driver can maintain the speed with the accelerator and take over the vehicle after the lane is recognized.

(Lane change function)

In addition, drivers can easily regain control of the system by lightly tapping the steering wheel in some merging scenarios where turn signals are not required to be activated. Unlike some products, the lane-keeping system engages in “competition” with the driver and increases torque to maintain the lane, and the driving assistance system will directly exit when the driver takes over the direction.

Execution Control Layer

The Baojun Kiwi EV has points to improve and do well in the execution control layer. It performs well in acceleration, which allows drivers to accelerate above the preset speed of the vehicle when the road ahead is clear. After releasing the accelerator pedal, the system returns control of the vehicle, and it slowly decelerates to the preset speed.The downside is that when the vehicle is automatically changing lanes, adding steering force manually will cause the vehicle to deviate from the target lane due to the superposition of the two torques. There is room for improvement in this aspect.

(After turning on the turn signal, the system automatically engages the lane change function. However, if the driver intervenes by turning the steering wheel, the two torques are superimposed and the vehicle deviates from the target lane. That's why you can see in the video that we manually counter-steered).

During this urban test drive, we also found a problem that needs to be solved. Apart from following scenarios, the assisted driving system of the Baojun KiWi EV cannot actively stop at a red light in the position of the leading car. In fact, it is difficult to achieve this with a vision-based assisted driving solution alone.

According to the product manager of Baojun, due to the non-uniform traffic light standards and locations in various cities in China, it will take some time for the visual algorithm to learn. However, Baojun is working hard to test this function, and the company has announced that it will be pushed to car owners via OTA in the near future.

Final Thoughts

At present, the price of the Baojun KiWi EV with the “Lingxi Smart Driving System” in 2023 has not been announced, but it is expected that its price will still be within the reach of the basic customer market.

Compared with most L2+ intelligent assisted driving products on the market, apart from the highway scenario’s NaviAssist function, the “Lingxi Smart Driving System” provides an almost unparalleled experience. It also reflects the rich thinking of product managers and engineers in urban driving scenarios through certain decisions.

This article is a translation by ChatGPT of a Chinese report from 42HOW. If you have any questions about it, please email bd@42how.com.