The meta-analysis quantitatively evaluates the technical characteristics (DoF, mass, COT, speed) and control methods (RL, ZMP, MPC) of humanoid robots. Using a random-effects model in RStudio, key metrics were established: a mean COT of 0.79, speed of 1.48 m/s, and a task success rate of 68.14%. The high efficiency of quasi-direct drives and reinforcement learning algorithms was confirmed. A classification of robots for medical, industrial, and domestic applications is proposed. The study systematizes data to optimize design and investment strategies through 2030, highlighting the need to enhance the autonomy and manipulation precision of anthropomorphic systems.
Aleksandr Shapurov (Mon,) studied this question.