After receiving the instruction, the engineer issued a voice wake-up command to the prototype.

"Start demonstration mode."

The LiDAR on top of the wheeled robot's body slowly rotated.

The sensors all over its body immediately entered working status.

It slowly advanced, moving back and forth on the simulated home floor, slowing down in advance when encountering scattered cardboard boxes and then bypassing them.

Even when encountering slightly lower steps, it could easily pass through.

Soon, it encountered a step with a relatively large height.

It extended two mechanical arms to support the vertical surface of the step, then applied force with the arm joints to lift the body, coordinating with the wheeled chassis to inch up the step.

It's just that the movement couldn't be considered smooth and flowing.

Moreover, there was a clear pause at every step because the algorithm needed to repeatedly calibrate the center of gravity.

The tester on the side was also staring intently at it.

Fortunately, it was a close call, but it passed smoothly.

The person in charge on the side also explained to Jiang Huaixing.

"President Jiang, currently we have deduced the scheme of mechanical arm suspension crossing suspended obstacles in the simulated environment.

But in multiple tests, it would overturn due to loss of control over the center of gravity.

Therefore, this function currently stays only in the laboratory and cannot be implemented yet.

..."

Subsequently, he continued to introduce the hardware configuration of this robot item by item.

"This set of wheeled mobile chassis it is equipped with is a derivative version of the industrial AMR platform.

Because it has already been verified by port and factory projects, it is already very mature in technology.

Then we also mounted the core control system of the housekeeping robot and two sets of degrees-of-freedom collaborative mechanical arms on top of the chassis.

The technology of this mechanical arm comes from industrial and catering service robots.

...

Although hardware like chassis and mechanical arms can be manufactured by many manufacturers.

But what truly widens the gap is the perception system, algorithm framework, and local computing power."

He pointed to the top of the robot and continued.

"Let's look at its perception system first. What is mounted on its top is our self-developed LiDAR, which realizes spatial perception by building 3D point cloud maps.

And multiple groups of sensors such as high-definition, infrared, and ultrasonic are also arranged on the body.

..."

"The most complex part is still its arms.

Its dual arms integrate our self-developed tactile sensing array.

This array has built-in thousands of micro-sensing units.

It can recognize pressure, friction, object softness and hardness, temperature, etc.

Coupled with the hand control algorithm,

it can imitate some simple human hand movements."

Speaking of this, the tester issued an instruction again.

Its mechanical arm was seen reaching out, then clamping a banana, and slowly completing the peeling action.

And the banana was not broken.

Don't look at how humans doing this seems very simple, but placed on a robot, it is actually a very difficult task.

It needs real-time feedback through sensors, then requires continuous calculation by algorithms, and finally can complete this series of actions.

"To make it perform this series of actions,

we have arranged a total of more than 280 various sensors internally.

The data collected by these sensors will be aggregated and transmitted to the intelligent control center.

Its main controller adopts the AI chip designed by Xinghan Semiconductor.

Although it was initially developed for White Star Auto automatic driving,

it belongs to a general-purpose AI inference chip.

Therefore, various unmanned equipment can be adapted.

Its single-chip peak computing power is 28 TOPS, and the whole machine is equipped with two pieces, totaling 56 TOPS of computing power.

Because the amount of tactile data from the hands is huge, we specifically customized an auxiliary chip with a 12-nanometer process from Xingchen Microelectronics, specifically used to process tactile sensing, thereby sharing the load of the main chip."

"This housekeeping robot can be said to integrate all kinds of advantages of our other robots currently.

Although it is positioned as a housekeeping robot, it does not affect at all that it is currently one of the devices with the highest software and hardware integration complexity for us.

It converges a large amount of our cutting-edge technological accumulation at the present stage."

During the introduction, this robot drove towards the dining table.

Jiang Huaixing watched it use its mechanical arms to tidy up dishes, and then dump the kitchen waste inside.

Finally, it sent the tableware into the dishwasher.

Although its movements were relatively slow throughout the process, and recognition lags even frequently occurred,

being able to complete this process was already very good.

After watching the demonstration, Jiang Huaixing asked,

"The thing is good, but how much does it cost, and what price do you plan to sell it for?"

Hearing this, the person in charge scratched his head.

"Uh, currently, our small-batch trial-produced BOM material cost is 180,000, and as for the selling price, we haven't determined it yet."

"Only calculated the hardware cost? Then R&D investment hasn't been amortized into it?"

Hearing this, Jiang Huaixing directly grasped the key point in his words.

"Uh, if all R&D is amortized evenly, then commercialization is almost unviable."

The person in charge said with slight embarrassment.

"Because the cost of direct R&D investment on our side is about 1.3 billion US dollars.

There are also chips and sensors, which belong to group-shared projects.

If their R&D costs are included, the total investment reaches more than 4 billion US dollars.

If all these are amortized into it, its selling price will be ridiculously high.

Therefore, at this stage, we only accounted for our direct investment, and other modules use group internal settlement."

Hearing this, Jiang Huaixing understood.

This housekeeping robot will inevitably be in a state of long-term loss after being listed on the market.

Because the R&D cost is too high.

Although the R&D expense of baixing mobile is even higher than this,

the problem is that baixing mobile has a shipment volume of about 100 million units per year.

the pressure of its huge R&D cost amortized to each mobile phone is very small.

And this thing called housekeeping robot looks like a niche product at first glance.

The initial shipment volume will definitely be extremely limited, so the amortization pressure will be very heavy.

Fortunately, the high profits of White Star Technology's auto and mobile phone products can continuously transfuse blood to the robot track.

Otherwise, Baixing Robot itself would have probably gone bankrupt long ago.

After visiting the housekeeping robot, the group came to the special robot testing room next door.

Speaking of special robots, they are actually robot dogs.

There are several forms of robot dogs displayed here:

some are traditional four-legged, some are six-legged, and there is also a wheel-foot composite form.

These robot dogs also each have their own advantages.

Especially the wheel-foot composite robot dog.

It has two forms.

One is a wheeled form that can move quickly on flat ground, and the other is a footed form on complex terrain.

"President Jiang, just last month, we just finished a test with our partner.

Our mechanical dog excellently completed their reconnaissance mission in the ruins simulation test.

At the same time, our six-legged robot dog also performed very well in load-bearing transportation under complex terrain conditions.

..."