Evolution of humanoid robots

 

Figure1: Evolution of humanoids

Introduction to humanoid robots:

None of humanity's creations inspire us with such an odd mix of terror and admiration as robots. Designing and working of humanoids was an intricacy in the past. There were a number of reasons for this, one of which is the myriad obstacles that science and technology faced, such as battery issues.

Professional service robots called humanoid robots are designed to move and interact in the same way as humans do. They add value, like all service robots, by automating duties in a way that saves money and increases productivity. Humanoid robots are a type of professional service robot that is still relatively young.

When working in human-populated areas, a humanoid robot has significant advantages. The key benefit is that a humanoid robot can act like humans in such a setting without any prior adjustments. Today, because of advances in AI, humanoid robots can do a wide range of functions, including successfully navigating their surroundings, detecting items in their environment, and assisting people with jobs such as bricklaying, drywall installation, and robotic-assisted surgery.

As a matter of fact, these advancements in technology as well as the combined development of the fields like mechanical, robotics, computational and artificial intelligence have come about over centuries of research and experimentation. In this blog, the evolution of the humanoid robot, from the very first idea of a humanoid robot in the past to its unfathomable possibilities in the future, has been discussed.

Past

1. Leonardo’s Mechanical Knight

Figure2: The mechanical knight design

Figure 3: The mechanical knight 

In 1495, Leonardo da Vinci, who is regarded as the first man, drew a humanoid mechanism. Unlike most of his inventions, Leonardo da Vinci appears to have produced the robotic knight, an actual knight's suit of armour with internal mechanisms. It was largely employed as a kind of entertainment at parties. The robot knight could stand, sit, lift its visor, and move its arms freely, and its jaw was anatomically perfect. A set of pulleys and wires controlled the complete robotic apparatus. The robot has been faithfully recreated based on Leonardo's design since the discovery of the sketchbook and has been found to be entirely functional.

2. Soldier with a trumpet

Figure 4: Soldier with a trumpet humanoid

The soldier with a trumpet, made in 1810 by Friedrich Kaufmann in Dresden, was the first humanoid robot. It is undoubtedly one of the most well-known androids that have survived to our time (it is now in the Deutsches Museum's collection, München). The Trumpet Player is a figure of a man clothed in traditional Spanish garb who is about 180 cm tall. It had leather bellows for lungs and reeds that sounded like a trumpet. It was even able to simultaneously blow two different tones. It had a special feature of playing two notes at the same time.

Trumpet Player's mechanism is controlled by two rotatably placed brass stepping drums. The notches on the drums make contact with six impacting tongues (pins) and four scanning levers, which activate wind valves that allow air to move through 12 tongues, resulting in a sound modified through a trumpet. The spring system (two helical springs) that powers the stepped drum and bellows must be wound up using the visible hand crank on the right.

3.    WABOT 1 and WABOT 2

Figure 5: WABOT 1 and WABOT 2

Wabot 1 was the first humanoid to walk on two legs, built in 1973 by Waseta University’s professor Ichiro Kato. It was the world's first full-scale anthropomorphic robot, with a limb control system, a vision system, and a built-in communications platform. Sensors allowed Wabot-1 to gauge distances and directions to objects, and the robot could walk and even grip and move objects with hands that used tactile sensors. Since then, robot technology has advanced significantly, despite the fact that it could only take a few shaky steps at the time. It has progressed in recent years to the point that robots can move away from obstacles or run on their own, indicating that humanoids have a bright future in the marketplace.

Following the construction of the WABOT-1 in the early 1970s and its sequel, the WABOT-2, in 1984, Waseda University and Hitachi demonstrated the WHL-11 in 1985. This bipedal robot stood 1.4 meters tall, weighed 120 kilograms, and had a leg length of 0.9 meters. Each leg had six degrees of freedom, with all 12 degrees of freedom being controlled electro-hydraulically.

4.    Honda P series humanoid robots.

Figure 6: The evolution of Honda humanoids

The P series is a sequential progression of Honda's prototype humanoid robots. As a result of the research, ASIMO, the world's most advanced humanoid robot at the time, was created.

•P1 developed in 1993

•P2 unveiled in 1996

•P3 unveiled in 1997

The technology needed to make these capabilities a reality was created.

1.High-level postural balancing capacity, allowing the robot to retain its stance by quickly extending its leg.

2.External recognition capability which facilitates the robot to include data such as people's movements around it

3.Ability to generate autonomous behaviour, allowing the robot to make predictions based on received data and select the next course of action while being managed by an operator.

Development of robot technology:

Robot vision: Between 1980 and 1990, the concept of robot vision was established. A robot is programmed to "see" as humans do and interact with objects through two processes in Robot Vision:

•Imaging: Scanning 2D/3D items through the cameras connected to the robot.

•Image Processing: Post imaging, the robot processes the image using AI/ML algorithms such as the boosted cascade classifiers, dictionary-based object detection algorithms, and Convolution Neural Networks.

•Imitation learning: In 1999, imitation learning was used to train robots, and it is currently included in reinforcement learning. It's comparable to observation learning, in which babies learn how to do things by watching adults do them. In imitation learning, a robot is trained to accomplish a project by learning a mapping between observations and actions via examples, which serve as the training set. To attain the final goal, it then imitates the activities seen in the demonstration.

Present:

Advances in processing power and speed, software, interfaces, sensing, and actuation, as well as a developing understanding of human biomechanics, catalyzed significant advancements that started around the year 2000.

Robotics as a scientific and technological subject is changing as a result of the development of humanoid robots. New sensors and microcomputers reduce the barriers to integrated control and dynamic systems. The following are some of the current Humanoid Robots:

1.    ASIMO:

Figure 7: Honda's ASIMO humanoid robot

Honda's Asimo, which stands for Advanced Step in Innovative Mobility, is a humanoid robot that debuted in 2000. The technical details of ASIMO are described in Table 1

•A visual and sensory system has been employed for navigation in ordinary situations and for human interaction. • An intelligent system consists of a Frame grabber and PC for image processing and speech recognition, a processor for control and planning, radio communication with the external system, and a DSP board for identifying sound sources. • On the head unit, two board colour cameras are mounted to generate stereo images, which are then processed to compute depth. • The vision computer is connected to the PCI bus through a frame grabber for high-speed data transfer. • Two microphones are mounted on the front side of the head for sound detection. • Ethernet is used to communicate between processes that run at various speeds, for as the sluggish vision process and the quick sensory system. • A map management system is included in ASIMO's external system for task navigation in places such as offices, museums, and hospitals. There's also a database of other people's faces to help them recognise them. • The planning system handles both fast and slow circumstances for obstacle avoidance. • A 2D gesture recognition algorithm uses a statistical model to recognise the position of the hand and estimate the action. It can recognise the hand, face, as well as the body's side and front profiles. It can also recognise gestures such as handshakes, hand circling, bye-byes, hand swings, and high hand, which may be found here.

Figure 8:2D gesture recognition of ASIMO

2.    Sophia:

Sophia is developed in 2017 by Hanson Robotics, a Hong Kong based company. She is a sophisticated humanoid robot capable of determining, maintaining, and identifying individuals.

Figure 9: Sophia Humanoid Robot

•Sophia's AI is composed of three main components: neural networks, machine perception, and cognitive architecture. relying on the truth These are coupled to necessitate actions •Sophia's machine perception component enables her to understand data in a manner that is similar to how we utilise our senses in the real world, allowing her to recognise faces. Sophia is characterised as a software platform, and she is divided into three programs: 1. A Timeline Editor – A speech-scripting software (such as RISE) that allows you to respond to pre-determined queries. This essentially allows Hanson Engineers to program what Sophia goes to mention during a given situation. 2.A Chatting System – Sophia determines an acceptable response to a situation based on the context, similar to how Alexa and Siri work. In comparison to the Timeline Editor, this requires a higher level of intellect, although it is still bound by the library of responses. 3.OpenCog – Sophia can now be told about her past experiences and reply appropriately, which aids her in answering reasoning and logic questions. •Sophia would be a good fit for healthcare, customer service, therapy, and education in the long run.

Robotic E-skin:

Electronic skin is a material that mimics the Human Skin in one or more ways and can sense pressure, temperature, and touch.

Figure 10: Robotic E-skin

Development of E-skin

2010: Attaching nanowire transistors to the sticky substrate, embedded in thin pressure-sensitive rubber capable of sensing a wide range of pressures.

2011: Stretchable solar cell used to power the electronic skin.

2012: Self-healing capacity, made of plastic and nickel.

2013 : An electronic skin that lights up when touched is developed.

Features of E-skin

1. It can detect heart electrical activity, brainwaves, and other vital signals. 2. It has the ability to record electrical activity on the scalp. 3. Neck muscle contractions can be controlled.

Structure of E-skin

Figure 11: Structure of E-skin

1. Nanotube TFT drain is connected to the anode of OLED.
2. OLED-: BI Layer structure whose colour is controlled by emissive layer material.
3. PSR: Electrical contact with cathode OLED.
4. Conductivity of PSR is directly proportional to applied pressure.

5. AMOLED: We can arrange Single-pixel OLED control circuitry into an active matrix.

Figure 12: AMOLED

Above figure represents a single colour flexible AMOLED display with 5 to 10 V.

Each pixel can be individually addressed using nanotubes.

Present Applications of humanoids:

The use of Humanoids was at an all-time high during the Covid-19 Pandemic. Wuhan Wuchang Smart Field Hospital was transformed into a robot-field hospital for patients, with the following features: •A thermometry system was utilised to check patients' temperatures and warn personnel of fever signs. •To guide the patients for hospital admission and to provide necessary information Humanoid robots were used. •Robots were utilised to provide food, and medicine, and disinfect the floor. To reduce the possibility of contamination, hospitals were heavily automated by humanoids. Humanoids were used in healthcare during the epidemic in the following ways: •Telepresence and mask detection: Hospitals deployed robots like Pepper to enforce social separation and mask wear. •Disinfection : Using UV radiation, the Xenex robot-assisted in disinfecting wards.

Future

Humanoid robots have the greatest potential to become the industrial tool of the future while being one of the smallest groups of service robots now on the market. Human-looking robots have been developed by companies such as Softbank Robotics for usage as medical aides and instructional aids. Humanoid robots are currently thriving in the medical field, particularly companion robots.

1)    The Tesla bot

Figure 13: Technical developments in the Tesla bot

The Tesla Bot is a 56-kilogram, 1.7-meter robot. The bot will deliver information on a screen placed where its face should be. The humanoid robot will be able to carry approximately 20 kg and deadlift around 68 kilos. 

Tesla's Autopilot software will be used by the bot. It will include eight cameras that will feed into a neural network developed by Tesla for its FSD software.

The neural network mimics human brain processes by allowing the vehicle to analyse its surroundings via cameras and determine what it needs to do when it faces barriers by identifying and labelling various pathways and images.

2.    NASA's Valkyrie humanoid robot

NASA's Valkyrie robot is being used for similar purposes, though on future Mars missions. Valkyrie is a 300-pound 6-2 humanoid robot. Two Intel Core i7 computers power the robot's brain, and lidar sensors, cameras, and a Multi-sensing SL camera are housed in the head to constantly scan the surrounding objects and environments. The Multi-sensing camera senses the environment using laser, 3D stereo, and video. Hazard cameras scan ahead and behind the torso for potential threats.

Figure 14: Valkyrie, the super-hero humanoid

A robot like Valkyrie's ultimate purpose is to man a future Mars trip. They can be deployed on scouting missions to the planet and utilised to establish living compounds, as well as provide power and life support for future manned missions.

Combinations of arm, wrist, finger, and thumb movements that collectively complete a task are still being developed, but they will make operations like tightening a bolt with a wrench in a circle or hauling a cart from one location to another possible.

References:

1) https://link.springer.com/chapter/10.1007/978-3-540-77296-5_5
2) https://www.cnet.com/pictures/the-evolution-of-humanoid-robots-pictures/6/
3) https://www.latentview.com/blog/the-rise-of-humanoid-robots/
4) https://en.wikipedia.org/wiki/Sophia
5) https://www.washingtonpost.com/technology/2021/08/19/tesla-ai-day-robot/
6) https://researchfdi.com/tesla-bot-elon-musk-robot/