The Synthetic Human Project
GRAND and software giant Autodesk are partners in a large-scale project to construct a true-to-life synthetic human.
Posted by GRAND NCE, March 13, 2014

The Synthius Project will put our collective science about humans into a single system. IMAGE: Wikipedia.
The Synthius Project aims to integrate scientific models about humans into a single system. IMAGE: Wikipedia.

(VANCOUVER) What is a human? What are all of the different facets that make up who we are? Though a complete answer to these questions would be impossibly complex, the more concrete problem, from the vantage point of science, considers how we might faithfully model what it is to be human.

Such a model has a good deal to account for. Our physical presence, our body, includes the full complexity of our anatomy, senses, and physiology, while our minds are likewise multipart – endowed with cognition, emotion, and memory. We are also social beings, with complicated interactions and behaviours that change according to the particular situation. All of these human aspects are moreover interconnected, and integral to the organism as a whole.

Many research projects across Canada and around the world have modeled the “ingredients” of being a human. None, however, has successfully put them together in an integrated model. Put differently, none has attempted to fully simulate the organic whole of a real human.

This is the goal of the Synthetic Human or “Synthius” Project, a new research project within GRAND and arguably its most ambitious one yet.

“Bringing it all together – the ‘synthesis’ – is a key component of the overall Synthius Project,” explained University of British Columbia Electrical and Computer Engineering professor Dr. Sidney Fels, who leads the project. “The connections are about all that it means to be human. It’s about finding out how research in other areas can be pulled together to develop a single integrated digital human system.”

Working in close collaboration with Autodesk, the multi-university Synthius project integrates six distinct research areas within GRAND, each modeling a core aspect of the human: physical form, motor control, biomechanics, cognition, emotion, and social behaviour. State-of-the-art robotics, artificial intelligence, sensor technology, and computational linguistics - to name a few fields - have already contributed to sophisticated models of each of these aspects. Their applications range from surgical simulation, to ergonomic design, to the design of believable game characters. They also serve as advanced tools for studies in areas such as human interaction, communication, and motor learning.

The Synthius Project focuses this advanced research on a common goal: the creation of a complete model of the human mind and body – a “synthetic human.” Each area, cutting-edge in its own right, will contribute to both physical (robotic) and computational models of the human body, as well as computational models of our social, cognitive, behavioural, and emotional characteristics.

“From the bones, to the muscles, to the nerves, to the brain, to the cognition, to the social and expressive human, all of those pieces will be [in the synthetic human] and predictable,” said Dr. Fels. “We’re going to make the whole human.”

Human data brought to life

“When I talk about the Synthius Project with other people, they say ‘that sounds really cool but isn’t somebody already working on that?’ The answer is no,” said Azam Khan, head of the Environment & Ergonomics Group at Autodesk Research in Toronto. “These amazing advances being done at labs across the country are great, but they’re so far from being useful to people because they really need to be integrated into a holistic system like Synthius.”

Khan and Autodesk Simulation Researcher Rhys Goldstein are Synthius Project Champions. Their role is to communicate customer or end user needs to the researchers and help guide the research objectives. Khan’s manager at Autodesk Research is Dr. Gord Kurtenbach, former chair of GRAND’s Research Management Committee.

For designers, engineers, and creative professionals – Autodesk’s primary customer base – access to robust models of human anatomy and behaviour will help them design better things for people.

“Let’s say you’re designing a cellphone and want to know how people can dial using one hand. You need to know about their hand size, and how long they can reach, how they would hold the phone,” said Khan. “If we had a digital human model, it would help the designer understand how humans work and how a design can make things better or worse.”

This crossing of human modeling and ergonomic design is the focus of the Synthius Project’s biomechanical division, also known as the Parametric Human Project (PHP). Founded by Khan, the research consortium of Autodesk Research, NSERC Canada, and FARO Technologies has set out to build a comprehensive database of the human being.

Surprisingly little is known about how human anatomy varies across age, sex, or geographical origins. Now, Canadian researchers are collecting data about muscle architecture and other structures of the human body – something done nowhere else in the world. By incorporating this “parametric data” into future software applications, designers will be able to dial in any age, height, or other characteristics to accurately model the intended users of their designs.

Just as Google created a database of the world with the now indispensible Google Maps, Dr. Fels sees the Synthius Project creating an indispensible database of the human, and similarly open to wide-ranging applications.

Dr. Fels’ has developed a simulation platform for modeling human anatomy that will become a cornerstone of the Synthius Project’s biomechanical model. Among other applications, the model will help doctors better plan their surgeries and simulate the effects of an operation across the entire body.

“You could go on to AutoCAD to piece together a skeleton and put some sort of animated muscle in there to model, say an operation on the jaw” said Dr. Fels. “But when you go to actually cut that muscle, what impact will that have on, for example, your bite force? Are you going to be able to chew again?” These are the kinds of questions that simulations on a synthetic human would help answer.

During his interview, Dr. Fels demonstrated a prototype cranium that is part of a physical model of the synthetic human. The robotic head revealed its intricate assemblage of joints, gears, and finely calibrated servomotors all framing two protruding mechanical eyes. As the head springs to life, the neck swivels and tilts and the articulated segments move in coordinated but random movements causing the face to twist and gesticulate.

“This is the lowest level of a synthetic human,” he explained. “You can build a robot head on its own, but unless you give it cognitive, expressive abilities with reasonable motor control, it just looks like a robot head moving around with no point to it.”

Such expressive synthetic human characters or “virtual actors” – either in physical or digital form – may join the casting call of feature films and video games. Animators would control the features, personality, and mood of the character whose movements and facial gestures would be decided by expressive and behavioural models built into the software. The time and effort needed to program each movement and gesture of animated characters would thus be minimized.

“Even a few years ago, the most advanced character at the time was Gollum in Lord of the Rings, which had literally hundreds of controls for movement. It was just unmanageable,” remarked Khan. “One of the people working on that project asked about a different way to animate these characters. What he was really asking for was something like a synthetic human.”

A similarly high-functioning synthetic human may also serve as an intelligent assistant or guide, a therapist for autistic children, or a companion for the elderly.

Synthetic humans may even help architects and designers design more efficient buildings.

“Autodesk software is used to design all kinds of buildings. We need to do better helping architects and designers design more efficient ones. A key part of that efficiency is doing much better at understanding the occupancy of buildings.”

As building plans often assume full occupancy all the time – a wasteful model – realistic simulations provide a more accurate picture and show how people actually use buildings and consume resources to inform the design of work and living spaces.

Given the cooling, heating, and operations of buildings is the biggest contributor to greenhouse gas emissions, even minor improvements to efficiency can translate into big impacts – especially given the near ubiquitous use of Autodesk design software.

“Autodesk is good at creating scalable, very large software systems. That’s an element most researchers don’t have the capacity or even the backgrounds to do,” said Khan.

“The exciting thing about Autodesk is that they share this vision,” added Dr. Fels. Some of the activities are going to create new business models and new products. The objectives are sufficiently large that the process of working towards them will lead to the advancement of human knowledge.”

GRAND’s “big bet”

Synthius is the first of GRAND’s “alliance projects”, introduced with the network’s newly restructured research program. These large-scale, high profile research collaborations are tightly coupled with industry partners with each kicking in significant co-investment. The scope and crosscutting nature of these projects is intended to make a significant impact on the Canadian digital media sector.

“It’s a ‘big bet’ because it’s like going to the moon,” was Khan’s comparison. “It’s a big complicated project and it feeds off the spirit of GRAND, which is about the network and having people working together.”

In comparison with other projects in GRAND, “alliance projects” will have far greater engagement with their receptor community or end users. Autodesk and Boeing, for example, plan to integrate Synthius Project models into their 3D modeling environments.

Khan sees Autodesk firmly behind the project because it brings together industry and universities to work on a large common goal.

“It is really hard to build a digital human,” said Khan. “GRAND is providing a great access to the research community in Canada. It provides Autodesk the infrastructure and support to even attempt to do something like this. It would be a lot of work to find out who are the key people in all these different areas.”

As a research platform, the Synthius Project will provide a “common vocabulary” for investigators to bridge different areas of the project and contribute to a shared synthetic human model. Given its scope and scale, Dr. Fels believes the project will greatly advance each component research area, but also achieve a human model that is impossible to realize in isolation.

“By bringing together people towards a common activity, you actually create something that is bigger than the sum of the parts,” said Dr. Fels. “By leveraging all of that expertise – that’s what Synthius will try to achieve.”




Media Contact:

Spencer Rose
Communications Officer