Focus on analysis-don't underestimate the warehouse forklift.

Edit | Peng Xiaoqiu

The first half of the year has passed, and there is little splash in the robot's pond.

Interestingly, however, a subdivision of the mobile robot track, which was previously considered boring and not optimistic, is in full swing in indifference. In the first half of the investment industry, unmanned forklifts have announced a number of financing: domestic brands including future robots, Jiecong, micro-robots and wood ant robots have successively completed 1 100 million yuan financing, with the highest amount reaching 500 million yuan; Foreign brands have Seegrid. According to 36Kr, there are other unmanned forklift companies that have completed financing, but they have not yet been announced.

Many times, players on the same track share weal and woe. But also in the name of "mobile robot", when the road of AMR financing is bumpy, why does the unmanned forklift usher in a small hot spot?

Forklift trucks, industrial vehicles used for loading and unloading, stacking, short-distance transportation and heavy load handling in warehouses, factories, airports and other places. It is called a forklift because there are often two fork legs in front of the car to lift the goods on the pallet. Unmanned forklift, that is, applying AGV technology to forklift, is the intelligence of traditional forklift. Technologies such as autonomous navigation, map algorithm and multi-vehicle scheduling system used by mobile stations such as AMR are also applied to unmanned forklifts.

The investment in unmanned storage has been carried out for at least five years, but under this theme, the popularity of unmanned forklift trucks has been far weaker than AGV and AMR. Compared with the latter, the target in the track of unmanned forklift is quite limited. Although AGV companies such as Jizhijia and Kuaicang also have unmanned forklift products; Traditional forklift enterprises such as Hangcha, Heli and Linde China are also introducing intelligent products. However, for companies that are mainly forklifts and still in the early stage of the primary market, the financing in the first half of the year has covered almost all the targets.

The reason behind the uneven heat and cold of AGV/AMR and unmanned forklift trucks may be that capital observes them from different logics.

After nearly two years of capital competition, AMR has to quickly match its high valuation with a large number of orders, and getting customers to accept new products is a slow process, which is also the current dilemma of AMR.

An unmanned forklift is not. For factories and warehouses, forklift trucks are just needed, and its application has almost continued the entire era of manufacturing development. This type of vehicle is familiar enough. At the same time, forklift trucks are widely used in factories, warehouses, airports and docks, and forklift trucks are needed in scenes that require stacking and loading and unloading actions. The application industry itself fluctuates again, and forklifts are all "the east is not bright and the west is bright".

In essence, forklift trucks, whether electrified or unmanned, are a mature upgrade process, and the psychological threshold for customers to accept is low. Therefore, the demand fluctuation frequency of unmanned forklift is the same as that of traditional forklift.

The forklift itself is in a boom cycle. During the ten years from 2009 to 20 19, China has become the largest forklift consumption market, with the growth rate exceeding that of the whole world. According to the development of the past decades, the sales volume of forklift trucks is similar to the curve of manufacturing industry. Every time it bottoms out, there will be a boom cycle of 2-3 years. According to the research of Zheshang Securities, domestic forklift sales reached a new low in 20 19, and will continue to increase in volume for three years from 2020 to 2023.

In short, after witnessing the self-certification dilemma of AMR, Capital hopes to find the goal of "replacing" rather than "creating demand" in the field of intelligent manufacturing of industrial vehicles, so as to improve the certainty.

A figure can prove the market demand for unmanned forklifts. According to the data of China Mobile Robot Industry Alliance (AGV/AMR), in 20021year, the domestic sales of fork mobile robots reached 8,000. This figure seems small, but since 20 17, the compound growth rate of forklift AGV sales has reached 78%. According to the logic of replacing forklift trucks mentioned above, there will be thousands to tens of thousands of vehicles for every increase in forklift truck ratio of 1% in the future. If this growth rate can be maintained, the market scale of unmanned forklifts will expand rapidly.

Compared with AMR mobile car replacing porters, it is more expensive for unmanned forklift to replace forklift drivers. The recruitment website shows that the average salary of forklift drivers is about 5300 yuan, and you need to obtain a forklift driver's license to get started. No matter from the recruitment difficulty or labor cost, the demand of customers will be more urgent.

To put it simply, the difficulty of unmanned forklift trucks is higher than that of AMR. Sun Jiao, who jumped into the capital, analyzed from 36Kr that unmanned forklifts are self-important, such as balancing heavy vehicles, the load is about 10 tons, and a car is only about10.5 tons; The running speed is fast, and the consequences are much more serious than AGV/AMR if it hits people/equipment.

The unmanned forklift is just not easy to control. As far as the car body is concerned, the motion model of unmanned forklift can also be abstracted as a bicycle model, just like unmanned driving. The steering system of this model belongs to "large lag system", so it can't turn around and change the driving direction anytime and anywhere. Therefore, when controlling its steering, it is necessary to consider the response of the vehicle in different cornering and speed States, which needs to be predicted and compensated in the software control system. In addition to the car, the unmanned forklift also accurately controls the execution unit-the fork. Especially in some advanced fork-taking scenes, the fork should ensure both positioning accuracy and safety.

In other words, the entrepreneurs of unmanned forklifts should not only understand the hardware structure, but also put these understandings into the software to meet the AGV demand of autonomous mobile. Higher intelligence and better handling are also the expectations of capital for unmanned forklifts.

Let me talk about intelligence first. This is the most essential opportunity for the unmanned forklift company. The top ten manual forklift companies in the world are giants, and seven of them are launching unmanned forklift products. However, their shortcomings are that they don't have intelligent R&D capabilities, and they all adopt third-party solutions, which leads to problems such as secondary development, high licensing fees, and unsuitable scenarios. However, although the current professional manufacturers of unmanned forklifts have advantages in this respect, there are still many internal strengths to be cultivated.

According to 36Kr, among the core positioning and navigation algorithms, some targets still use the third-party positioning algorithm; Some targets advertise themselves as visual navigation algorithms, but actually they mainly use the previous generation QR code navigation method. These can only be called improved schemes, which still have a strong dependence on the environment and have no advantage compared with the real lidar scheme.

Secondly, as a widely used vehicle in workshop warehouse, forklift trucks need to stop at many links compared with mobile cars, which puts high demands on the communication ability of unmanned forklift trucks. 36Kr once saw in a factory he visited that 10 unmanned forklift had to dock more than 20 production equipment, 4 warehouses and 4 rolling doors. This requires that the scheduling algorithm itself can pass the design of road network and fault-tolerant mechanism. In addition to the optimization of algorithm model, we should combine the understanding of factory production mode, production rhythm and production process with scheduling algorithm.

As far as machine engineering is concerned, there are more problems to be solved in landing. An unmanned forklift founder told 36Kr that some companies whose main business is not forklifts can quickly launch forklift products by buying traditional forklifts for unmanned use, but traditional forklifts and intelligent forklifts are completely different logics.

The traditional forklift is highly dependent on the driver's perception and decision-making ability, and the motion control accuracy of the body is very poor. Just with the changes of positioning and navigation, it is impossible to be as agile and intelligent as people. At the same time, because the traditional forklift does not reserve a full view for the camera, it often produces blind spots when reversing and turning, so the simple and rude combination can easily cause safety problems.