PNDbotics’ Humanoid Robot Adam Reportedly Climbs Onto a 1-Meter Box

humanoid.guide reported on 2026-10-15 that PNDbotics’ full-size humanoid robot Adam can climb onto a 1-meter box. The report cites a LinkedIn post by Space and Technology, not an announcement from PNDbotics itself. The post describes the robot analyzing the obstacle and planning its movement before placing one foot on the box and climbing up while shifting its weight.

If the description of climbing onto a 1-meter box is accurate, it requires a higher level of physical control than walking on flat ground. It could therefore shift the criteria for evaluating humanoids from whether they can move to how well they can operate in environments with elevation changes. However, the report is based on a LinkedIn post by Space and Technology, not an announcement from PNDbotics itself, so any assessment must take that into account.

The key point is that Adam’s movement was introduced through a post from an outside account, rather than as a capability officially promoted by PNDbotics. The description of climbing onto a 1-meter box involves more than simple walking or stepping over an obstacle; it presents the movement as a whole-body coordination capability involving obstacle recognition, motion planning, foot placement and weight shifting. However, the humanoid.guide article does not specify the control method and is not primary-source evidence supporting the details of a real-world demonstration, so the conclusions that can be drawn are limited. Compared with our related article, PNDbotics Exhibits Humanoid Robots for Factories at the Smart China Expo, the public picture of PNDbotics appears to be shifting from exhibitions toward the promotion of movement capabilities. However, both articles are based on publicly available information and do not show whether the products are in mass production or have been deployed by customers. What can be confirmed here is only that discussion of the company’s humanoid robots has broadened from factory exhibitions to their physical ability to overcome obstacles. At the industry level, this kind of reporting suggests that the criteria for evaluating humanoids are moving beyond appearance and demonstrations designed to impress toward practical tasks such as handling elevation changes and shifting loads. What matters in practice, however, is which control system can reproduce the same movement, whether it is repeatable, and whether its behavior changes across indoor and outdoor settings or different floor surfaces. The next questions to examine are whether movements at the 1-meter level can be reproduced consistently, whether they are remotely operated or autonomously controlled, and whether there is evidence beyond publicly available videos and posts.

For Japanese companies, a key question may be how well humanoid robots can handle features of factory environments such as elevation changes, carts and cardboard boxes. However, this report does not reveal the control method or repeatability, and additional primary-source information would be needed to assess the potential for adoption in Japan.