Lethal Shooter and a robot shooter: technology meets basketball

Exactly. Missing shots, or not being able to move or reach something. We struggled quite a bit with those more basic issues.

We haven't tested it exactly, but it can shoot from as far as the opposite three-point line.

It wasn't intentional, but as we wanted to do more and more things, it gradually got bigger – simply because of the weight of the hardware or the addition of motors. Ideally, the lighter, the better, so the fact that it's getting heavier is not really a good thing.

We changed the overall design a bit. Until now, we've been working on a strange concept where various blocks with their own will gather to form a humanoid and play basketball, and the costume hasn't changed much. But for Cue6, we changed the silhouette a bit and changed the surface a bit to give a sense of evolution, such as having different colours visible when the movable parts move in 3D knit. We've also put a bit of effort into the design aspect.

No, we didn't. We once asked Alex Kirk to provide us with videos to try and recreate the same movements under the notion that it might improve if we referenced human movements. However, the result was different because the physical structure is different. The robot couldn't move like a human. For example, too much torque was applied to the knees, and it couldn't maintain the posture. Since the robot can't move like a human, it ended up finding the optimal way for its own body to shoot.

For instance, the robot has a map in its head for moving around and constantly grasps its current location on that relative map. This technology is essentially similar to that of AGVs and self-driving cars. The motor movement and batteries use cells from racing car batteries, and some parts that are also used in cars have been repurposed. We also obtained assistance from people involved in car manufacturing and equipment development for parts processing. In the technology development process, we've integrated a lot of knowledge gained from car manufacturing.

While there was overlap with previous models, and we started development on Cue6 earlier, it took us about a year and a half.

The time has been increasing little by little. Cue1 and Cue2 took about half a year from the point we decided to do it. As the difficulty level rises, so does the time required.

I can't disclose the amount, but it might not be as much as you'd imagine.

There are just so many things robots can't do. At every challenge, we appreciate the amazing capabilities of humans. The number one thing is probably flexibility. Humans, in a sense, can do anything, but robots can only do certain things at present. That, I feel, is the biggest difference between robots and humans. I think we can handle things like making it shoot more accurately to the centre of the goal, but what about when the ball changes, or the location changes, or while running, or when the ball is moving? Adapting to various scenarios for each and every case is still quite challenging for robots. Perhaps the quickest way to improve would be to learn from a great coach like Lethal Shooter.

Strengths include precision, accuracy, the quality of repetitive motions and a lack of mental distress. Honestly, there aren't many advantages to robots. Things like the resilience of the ball, the condition of the ball and the venue, the temperature and the relationship with the people around us are always different. I've realised again how incredible it is that humans consistently make the best decisions in these varying situations.

I heard that there was a goal specially prepared for the showdown, so I prepared things like shots with different trajectories for the match.

Next, I'd like to make it capable of running.

We have to change the legs, so if we want to make it run, the things we've been able to do so far would have to be redone. For instance, with shooting, Cue uses a chest camera to determine the goal's position. Since its legs are wheels, it basically faces the goal directly and just measures the distance. Whether it's at 0 degrees or 45 degrees, once you've aligned the angle, you just measure the distance and consider how to move the body to throw that distance. But if it becomes bipedal, it can't always face the goal directly. The way of throwing changes, and if the legs go up and down, it's not just about understanding the horizontal direction but also the vertical one. There are quite a few things that can be done with the current Cue that would not be possible in the current state, so that part presents a major technical challenge. It would be like creating an entirely new robot.

I don't think so. Even if AI advances and becomes stronger than humans, there will still be human competition. I think robots and humans will exist as separate entities in their own worlds.