Listen to this article
In August, 2021, Shenzhen-based L4 autonomous driving technology company, Whale Dynamic, and Innoviz Technologies, a leading manufacturer of LiDAR sensors and perception software announced that they’re joining hands. What this means is that Innoviz Technologies will bring its solid-state LiDAR technology to Whale Dynamic’s next-generation intelligent autonomous driving (AD) platform.
Keen to know more about the company, its intelligent platform and the recent collaboration with Innoviz, Auto Futures caught up with Yufei Chang, Founder and CEO of Whale Dynamic.
Founded in 2018, Whale Dynamic’s initial focus was developing intelligent traffic applications by applying a novel AI algorithm with LiDAR and camera fusion. The products were all about software and customised hardware integration, mainly installed on the roadside, highways, tunnels, etc.
“In early 2019, we launched our autonomous vehicle development scheme. We believe our patented early-stage sensor fusion technique and LiDAR enhanced high-definition mapping tool would make a big difference in the AV industry,” says Chang.
“Now, our goal is to create a reliable autonomous delivery vehicle that could carry out the daily delivery workload in suburban areas. The five-year plan is to shift from non-passenger vehicles to passenger cars like robotaxi, robobus or completely new forms of mobility we define. Our ultimate goal, which might sound ambitious, is to implement the autonomous technology in other terrains like underground, subsea or even in orbit,” he adds.
Pushing the Boundaries
There are many players in the autonomous mobility space, and quite a few of them are in China. I was eager to know more on the technology powering Whale Dynamic and what sets the company apart from other players in the space.
“There are four major technological advancements. The core part which powers our AV is the multi-sensor fusion with early-stage synchronisation. The data from LiDAR, camera, mmRadar and GNSS/IMU are synched in spatial and temporal dimensions. The LiDAR enhanced mapping and localisation is another core technique coupled with our perception module. Therefore, our perception and localisation are mutually reinforced as a closed loop. The third advancement is our hardware design. To achieve higher sensor integration, we designed the whole vehicle roof-top sensor unit. From high volume data compression to thermal ventilation, from dynamic power distribution to rainwater resistance, we pushed forward the boundary of AV stability,” explains Chang.
“The fourth is we designed a control-by-wire chassis. This allows us to thoroughly apply our engineering capability and creativity to the vehicle we designed. For instance, most AV companies now use Lincoln MKZ, and it costs significantly to do the retrofitting; this type of design process works with efficiency at the very beginning as it was impossible for a software company to build up a passenger car.
“However, once the AV steps into the commercialisation phase, we have to take the vehicle cost, the degree of freedom that the automobile maker releases to us including control protocol and HMI, all into account. We could define the control-by-wire protocol and design the chassis accordingly. This considerable advantage plays as the second stage of rocket booster, which powers us directly to the commercialisation phase, faster than any retrofitting companies in China,” he says.
“The front and end direction are no longer necessary for robotics.”
Whale Dynamic’s solutions have been adopted by tier 1 companies like ADAS manufacturers and suppliers, EV makers and taxi operators, as well as Internet companies and universities. The company has been helping clients build the vehicle, consulting with them and sharing technological know-how along the way.
Apart from the obvious commercial benefits, the valuable experiences and feedback from clients also help Whale Dynamic further its own autonomous delivery vehicle projects. That said, most of these products are primarily in the R&D phase.
The company also recently lifted the covers off its unmanned logistics vehicle.
“When we designed this autonomous delivery vehicle, it was designated to be a Level 4+ delivery vehicle with the capability to run freely on urban roads. This vehicle is unique in two ways. One is the development process, or the methodology of development. We retrofitted several SUVs, put identical sensor configurations on top of them, and did the road-test day and night to verify its engineering stacks as aforementioned from perception to HD maps.
“Then we shifted to phase two, designed our chassis and migrated all that hardware and software from phase one to this chassis with the steering wheel and paddles still on it. This gives us the maximum degree of freedom to test our designed chassis. Phase three is what we are doing now, which consists of removing the steering wheel and paddles and letting the vehicle operate independently. Meanwhile, our SUV fleets are testing and propagating some advanced modules on passenger vehicles, and will shift these modules to the ‘unmanned vehicle’ once the result meets our design criteria,” says Chang.
“I’d also like to highlight the design of the vehicle. As you can see, it is symmetrical in left and right and in front and end. We don’t think the robotic vehicle has to follow the driving routing like humans. For instance, parking does not have to be front in and end out. The front and end direction are no longer necessary for robotics. Therefore, we make this vehicle bi-directional and give all four wheels steering functions. This design also saved a lot of production cost as we only had to produce a mould for one direction. This concept can only be done on autonomous driving.”
Whale’s Plans to Get Bigger
Our conversation turned to the collaboration with Innoviz Technologies.
“Whale Dynamic is keen to integrate Innoviz’s most advanced MEMS LiDAR on our L4 autonomous driving solution. If the development goes smoothly as expected, we will build a robust sensor fusion kit for L4 and above autonomous driving. The current solution heavily relied on mechanical LiDAR for which the price is very high.
“But the MEMS LiDAR has a small FOV (Field of View) that limits the application, and so we will put MEMS LiDAR together with our vision and mmRadar and some low-cost mechanical LiDARs to create an ultra-robust sensor fusion solution. We believe this is a more practical way to propagate the L4 technology than relying on single sensors and waiting for costs down. The cooperation is meaningful to the industry.”
“We are looking forward to becoming a listed company within two years from now. And shift our main business active region from China mainland to the US and EU after we validated the global business scope for autonomous driving (or, more specifically, for autonomous delivery,” adds Chang, speaking about Whale Dynamic’s future plans.
For Chang, rhe conclusion is clear. “The US is the best for both market size, public acceptance and legislation. The EU is good for market size and public acceptance, but a bit lagging in the legislation. So undoubtedly, we are going to these markets.”