Zhiji: The last battle before autonomous driving

When talking about smart driving, the first things that come to mind are Huawei and Xiaopeng. Since Zhiji Auto is a rising star, it is undoubtedly considered by most people to not be in the first echelon.

Yesterday, at the "Last Battle Before Autonomous Driving " open class held by Zhiji Automobile at Tsinghua Science and Technology Park , it presented to the industry a series of high-quality technical achievements that have been quietly cultivated in intelligent driving, which are still very good.

Most people here may want to ask, does Zhiji still have autonomous driving? I have experienced this before. Some time ago, Momenta, a partner of Zhiji, continued to invite internal personnel to experience test their latest system. In fact, it feels completely different if you don’t experience it physically.

In the technical field, algorithm optimization is what Zhiji has been doing. The core of the intelligent driving system lies in efficient algorithms that can analyze complex traffic conditions, predict behaviors and make safe driving decisions in real time. Through in-depth research and innovation, Zhiji has successfully reduced computing power requirements by 90%, which means that the system can achieve better performance with lower hardware resources. Through technical means such as model fixation, model compression, core algorithm automatic tuning, and cross-layer fusion, Zhiji has achieved a substantial improvement in model operating efficiency. It has not only made breakthroughs in functional growth, but also created new innovations in reducing computing costs. possible.

By being compatible with Xavier and Orin dual intelligent driving computing platforms, Zhiji has achieved the goal of efficiently running algorithms on different computing power platforms. It not only improves the adaptability of the intelligent driving system, but also ensures efficient performance in different scenarios. By optimizing the medium and low computing power platform, Zhiji IM AD has successfully achieved full-field visual fusion and can still maintain excellent performance levels even in complex environments.

In terms of AI technology, the DLP artificial intelligence model jointly developed by Zhiji Auto and Momenta provides an unprecedented solution to the decision-making and planning problems of autonomous driving. This model can more accurately analyze traffic conditions and make smarter driving decisions, thereby improving driving safety and efficiency. The introduction of the DDLD perception algorithm reduces the dependence on high-precision maps, achieves breakthroughs in full-scenario intelligent driving, and brings new possibilities for improving the driving experience.

IM AD is 3.2 times safer than human driving, significantly improving the safety standards of autonomous driving.

In the field of driving and parking, IM AD performs well, with a lane change success rate of 98%, safe takeovers of only 0.36 times/1,000 kilometers, and a parking success rate of 97%.

I have already experienced Zhiji IM AD, and it is indeed a big surprise in a real road environment. In many scenarios, Zhiji's automatic assisted driving performance is very good, and it is in the first echelon of automatic assisted driving in China.

From high-precision map NOA to urban NOA, to commuting mode in hundreds of cities, and finally reaching the "Door to Door" era of all-scenario commuting, IM AD will realize autonomous driving in most scenarios within 3-5 years. This player who everyone seems to be relatively backward has actually moved to the front through continuous efforts.

If you have the chance, you must give it a try.