HUMANOID ROBOT MODEL

A humanoid robot model is a robot that resembles the human body in form, function and actions.  It is typically characterized with a head, torso, two arms and two legs. It is designed to operate in environment and use tools built for humans.

There are two main types of humanoid robot model and they are; full body and semi/partial humanoid robots.

Full body humanoid robots are characterized physically of a complete human body structure with a head, torso, two arms, and two legs exhibiting bipedal locomotion. Example of full body humanoid robot includes; Boston dynamics Atlas, Tesla Optimus, Figure 03 etc.

Semi-humanoid robots have some but not all the physical human body features exhibited by full body humanoid robots. They may be legless or wheeled or built as torso/arm only on a fixed stand. Example of semi-humanoid robot includes; Diligent robotics Moxi, Amazon digit, PAL robotics TIAGO, Care-E, Nasa’s Robonaut, Rethink robotics Baxter and Sawyer etc.

The component of a humanoid robot are; structural frame (Skeleton), joints and degrees of freedom (DOF), sensors (perception system), power system, computing and control system, end effectors (hands/grippers), locomotion and balance control system, communication systems and outer covering/skin.

The skeleton or structural frame of a humanoid robot provides structural support and acts as the active linkages for the robots movements; as exhibited by the human skeletal frame. The skeleton is made of lightweight and strong materials such as aluminum, titanium or carbon fiber.

The actuator of a humanoid robot mimics the muscles of the human body. The actuator powers the joints and enables movement. The common types of actuators used in humanoid robots include electric motors and gearbox which are usually DC motors paired with harmonic drives or planetary gear boxes for torque multiplication, servo motors to provide precise control at a high torque, pneumatic actuators which uses compressed air to move the joints or hydraulic actuators which use fluid pressure to exert an extremely large force in the cases where it’s needed. Artificial muscles are increasingly used to actuate humanoid robots, they are made of electro active polymers, liquid crystal elastomer or fiber based myofiber actuators that contracts like biological muscles.

The joints of a humanoid robot consist or comprise of multiple joints that enables movement and flexibility. Theses joints are designed to exactly mimic human joints, such as the hip, knee, ankle, shoulder, elbow and wrist. The typical joint of a full body humanoid robot consist of several degree of freedom (DOF).  Where one degree of freedom is one independent axis of rotation/motion a joint can perform. For instance the allocation of joints in a full body humanoid is approximately 60 DOF.The more the degree of freedom the more the humanoid robot is closer to the human range of motion but exponentially and complex harder control algorithms required and more potential failure points.

The sensors of a humanoid robot comprise typically of the following sensors; vision, IMU, force/torque, tactile/pressure, encoders, microphones and other sensor types.

The vision sensor is usually stereo cameras or depth cameras (RGB-D) for 3D scene understanding. However some robots add LIDAR for precise distance mapping.

The inertia measurement unit (IMU) is accelerometers and gyroscopes usually placed in the torso/pelvis and the acceleration data gathered from them are essential for balance of the robot.

The force/torque sensors are placed at wrist, ankles and sometimes each joints to measure how much force is being applied or received, this is critical for safe manipulation and detecting ground contact while walking.

The tactile/pressure sensor are embedded in the finger tips and sometimes feet, sensing contact, slip and pressure distribution hence mimicking skin sensation.

The encoders are positioned inside each joints tracking precise angle/position of every actuator (proprioception).

The microphone is incorporated in the robot for voice command input or output and environmental audio awareness.

The power system of the humanoid robot consists of a battery, power distribution unit (PDU), thermal management system or tethered alternatives.

The battery chemistry of choice of humanoid robots is lithium. The advantages of using lithium-ion or lithium polymer batteries are its high energy density, long cycle life and relatively low self-discharge rate.

The power distribution unit (PDU) of a humanoid robot consists of devices that regulates and routes the voltage/current to numerous motors, sensors and the onboard computers, such as voltage regulators, power converters and other electrical connectors.

The thermal management unit comprises of heat sinks, thermal interfaces, cooling systems etc. whose main role is to control the heat dissipated by the numerous devices to an acceptable level required for smooth and efficient operation of the humanoid robot.

The tethered alternatives make allowances for the humanoid robots to be powered from external source such as the mains or generators.

The computing and control system of a humanoid robot is responsible for processing sensory data, making decisions and controlling the robot movements and actions. Generally the computing system of most humanoid robots has a hierarchical architecture with multiple layers of processing and control, while in most cases the control algorithms used to control the robots movements and actions are based on the mathematical models of the robot dynamics.

The end effectors of a humanoid robot are the components at the end of the robots limbs that interact with the environment such as hands, feet or grippers. The end effector of the hands are of four types; simple grippers with two or three rigid fingers, multi-fingered dexterous hands with 4 to 5 fingers with independent joints enabling fine manipulation, under actuated hands with fewer motors than joints using mechanical linkage/tendons to approximate natural fingers, and sensing integrated hands with the finger tips integrated with force/tactile sensors for grip control for tasks that require delicate handling.

The end effector for the feet of the humanoid robot consist of five main types; the flat rigid plate feet which are simple flat pads used for simple bipedal robots, the passive compliant feet are rigid plate with a springy or rubber layer underneath to absorb impact, the articulated/multi-segmented feet with a toe joint allowing the foot to roll through a step, the point contact/ball feet which are small rounded contact points rather than flat soles used in some dynamic or running focused robots and the wheeled feet exhibited by some semi-humanoids.

The locomotion and balance control system of a humanoid robot continuously fights gravity and momentum to keep the robot from falling, while also making it walk. It does this by employing the following techniques; balance control loops, gait generation, push recovery, zero moment point (ZMP) or model predictive control (MPC) and terrain adaptation.

The balance control loop technique runs a balance loop algorithm continuously (hundreds of times per second) using IMU and joint encoder data to calculate a center of mass and adjust joint torques to keep it over the base of support.

The gait generation technique is an algorithm of pre-planned or learned walking pattern usually obtained by reinforced learning or model predictive control. It defines how legs swing and feet place during each step.

The push recovery technique is an algorithm that compensates by fast reflexive adjustments (stepping, arms swinging) when the robot is bumped or loses footing.

The zero moment point (ZMP) or model predictive control (MPC) technique is a control framework used to keep the robot dynamically stable while walking, running or standing on uneven ground.

The terrain adaptation technique is an algorithm that uses sensors and a control regime to adjust step height, angle and timing for stairs, slopes or obstacles.

The communication system of a humanoid robot enables the robot to interact with its environment, humans and other robots. It allows the robot to receive commands, send data and exchange information with other devices or systems using devices such as Wi-Fi/5G/Bluetooth modules, tele-operation links, inter-robot /fleet communication devices and a safety communication channel.

The Wi-Fi/5G/Bluetooth module is used for remote monitoring, software updates, fleet management or streaming sensor data to the cloud.

The tele-operation links is used in some robot by a human operator to remotely control or guide the robot.

The inter-robot /fleet communication devices are used in robots deployed for a warehouse or factory setting. The robots use the communication channel to share task and location data with a central coordination system.

The safety communication channel employ emergency stop signals and safety monitoring algorithms on separate, highly reliable communication paths from general data traffic.

The outer covering of a humanoid robot is typically made up of protective shells, aesthetic skins or android style synthetic skins. The protective shell is usually made of plastic or composite panels. The aesthetic skins are soft, rounded covers such as synthetic leather and the android style synthetic skins are silicone based skins with embedded micro actuators to create facial expressions.

The advantages of humanoid robots are; humanoid robots can interact with humans in a more natural and intuitive way enabling more effective communication and collaboration. Humanoid robots can perform a wide range of tasks, from simple manipulation to complex tasks that require balance and coordination. Humanoid robots can be designed to operate safely in human environments, reducing the risk of accidents and injuries to humans. Humanoid robots can be designed to recognize and respond to human emotions, enabling more empathetic and personalized interactions.

The disadvantages of humanoid robots are; developing humanoid robots can be expensive, requiring significant investment in software, hardware and testing. Humanoid robots are complex systems that require advanced sensors, actuators and AI systems which can be challenging to design and integrate. Humanoid robots may be more prone to damage, wear and tear, particularly if they are designed to operate in dynamic environments. Humanoid robots raise ethical concerns such as the potential for job displacement, privacy risks and biases in AI decision making.

Humanoid robots will find applications in the following; humanoid robots can be used in retail, hospitality and health care settings to provide customer service and support. Humanoid robots can be used in search and rescue operations such as disaster response and recovery. Humanoid robots can be used in health care settings to assist with patient care, rehabilitation and therapy. Humanoid robots can be used in educational settings to teach children about robotics, AI, STEM concept and whatever other subject that catches the teacher fancy.

The future of humanoid robot is based on the advances and development of the following technologies; advances in AI and robotics will enable humanoid robot to become more sophisticated and capable with improved perception, manipulation and interaction abilities. Humanoid robots are expected to become more widely adopted in various industries including health care, customer services and education. Advances in manufacturing and design will make humanoid robots more cost effective, enabling wider adoption and deployment. Humanoid robots will enable new applications such as personal assistance, home care and entertainment and so on.

 

SOURCES:

• Humanoid robotics: An introduction by Katsumi Yamazaki and Tetsushi Ikeda.
• Control systems for humanoid robots by Jung-Hoon Kim and Frank C. Park.
• Humanoid robot vision and perception by Matteo Matteucci and Davide Scaramuccia.
• Introduction to humanoid robotics by Shuuji Kajita, Hirohisa Hirukawa, Kensuke Harada and Kazuhito Yoku.
• Humanoid robots by Dimitar Nenchev et al.