Human-Sized Bipedal Humanoid Robot

Starting in 2000, “Somjook” emerged as Thailand’s first and sole human-sized humanoid robot (though initially only the lower body/legs). Weighing over 50 kilograms, it was developed by the Field Robotics Institute as a prototype to study bipedal locomotion, featuring a total of 12 degrees of freedom. Somjook underwent continuous development, eventually achieving self-stabilized walking. However, further advanced development faced challenges due to the significant weight of its components, slowing progress. The research team thus decided to shift focus to developing smaller humanoid robots for ease of study and research, with the intention of applying the acquired knowledge back to Somjook later.

Small Bipedal Robot Prototypes

In 2003, the first small humanoid robot, Generation 1, was designed by the researchers. This bipedal robot stood 50 cm tall, weighed approximately 2 kg, and had 14 degrees of freedom. To keep it lightweight, its structure was made from nylon and aluminum alloy. It used R/C servomotors, similar to those commonly found in remote-controlled airplanes. The robot was equipped with two types of sensors: 16 pressure sensors to measure forces on both feet, and a 3-axis inclinometer to measure body tilt.

Jaidee: Small Bipedal Robot

By 2005, with the international World Robocup competition introducing a humanoid robot soccer league, “Jaidee,” the Generation 2 small humanoid robot, was created for the Robocup Soccer (Humanoid league) competition. In 2005, Jaidee made a remarkable debut by reaching the Round of 16. It subsequently improved significantly, securing a 3rd runner-up prize at the competition in Bremen, Germany, in 2006. Jaidee was a small humanoid robot weighing 3.3 kg, standing 38 cm tall, and capable of walking at 2 meters per minute. It could plan foot placements to dodge obstacles and approach the ball for kicking. Furthermore, if it fell, Jaidee could stand up on its own. Its walking speed could reach 1 meter/minute.

“Jeet” and “Kalamae”: Small Bipedal Robots

“Jeet” and “Kalamae” represent the Generation 3 small humanoid robots developed by the institute in 2007. This third generation featured high walking agility and enhanced impact resistance, making them ideal for robot soccer competitions. These robots were designed with the ability to maintain balance constantly, resulting in high walking stability and making them less prone to falling than older models. They could achieve a walking speed of 15 meters/minute.

PorDee: Small Bipedal Robot

The Generation 4 robot saw significant performance improvements. In the vision system, the camera’s light reception was switched from CMOS to CCD for improved color clarity and increased resolution, allowing for greater viewing distance compared to the older camera system. The processing system transitioned from a Microcontroller to a PC-104 with Windows XP Embedded, offering greater flexibility for connecting new hardware to the robot. Additionally, the operating system facilitated faster and more convenient code writing. In the locomotion system, a new knee joint design incorporated two motors per knee to increase knee-opening speed, resulting in faster walking.

Phoenix: Small Bipedal Robot

The Phoenix series represents the Generation 6 humanoid robot. Its structural design was modified to incorporate 19 degrees of freedom, achieved by reducing unnecessary degrees of freedom in the arms and adding a waist joint to enhance balance and standing capabilities. The size of the ankle motors was also increased to handle more torque during walking, and the main processing unit was relocated to the center of the body. This relocation aimed to reduce motor load and minimize damage from potential falls, resulting in an increased average walking speed of approximately 0.3 meters per second. For image capture, the camera was changed from an off-center lens position to a center-body position to improve image processing speed. The robot’s control program also saw adjustments to some constants based on the robot’s changed size, improvements in ball and opponent detection during gameplay, and modifications to ball approach methods for increased speed and accuracy.

Thailand Humanoid Robot Soccer Championship

The Thailand Humanoid Robot Soccer Championship, organized by the Thai Robotics Society with support from Seagate Technology (Thailand) Ltd., is considered one of the most challenging robot competitions. Participants must develop human-like robots capable of bipedal walking and possessing soccer game skills. The robots must operate autonomously without human control, differentiate between teammates and opponents, and collaborate as a team using only onboard processing units, without external processing. Humanoid robots are developed in both hardware and software, constrained by the size and weight limits of the components they must carry.

In the small humanoid robot competition, robots must be human-like in appearance and stand between 30-60 centimeters tall. Developing humanoid robots is more complex than other robot types because they typically have 20-22 motors in their joints for movement. Humanoid robots place significant emphasis on their walking system because having two legs necessitates the robot constantly maintain balance to ensure stability, whether moving or stationary.

Another crucial system in humanoid robots is the vision system. Humanoid robots must be able to perceive the location of the ball on the field, the goal, their own position, and opposing players through the camera mounted on the robot’s head.

A humanoid robot’s basic operational steps are as follows: the robot will first orient its camera to scan the field for objects. The image data from the camera is then sent for image processing to identify objects of interest, which in this case are the ball on the field and the opponent’s goal. After that, the robot’s main processing system calculates distances and commands the motors to drive the robot to walk towards the ball and kick it into the opponent’s goal, or perform other actions depending on the robot’s own game strategy, such as dribbling the ball to evade opponents or passing to a teammate. The robot’s intelligent decision-making system operates according to the program designed by the developers.