The concept of digital avatars has its roots in the early days of the internet. They were initially used to communicate on forums and later on social networks, employing images to represent themselves in the digital world.
Avatar technology has evolved over the last decade or so to become more human-like and realistic, which has opened up valuable use cases across many industries.
One notable example is Zoe, introduced in 2013 by Toshiba’s Cambridge Research Lab and the University of Cambridge’s Department of Engineering. Zoe was a virtual talking head capable of expressing human emotions such as happiness and anger, setting a new standard for avatar realism.
Digital Avatars In Wellness And Healthcare
According to GrandView Research, the global digital avatar market was valued at $14.34 billion in 2022 and is expected to reach $270.61 billion by 2023. I believe the wellness industry, in particular, has the potential to capture a significant amount of this market.
First and foremost, avatars can serve as companions. For example, Replika has developed an AI-powered platform that enables users to create personalized avatars to help users with “personal growth, professional fulfillment and mental well-being,” according to a recent Forbes article. Essentially, these avatars act as virtual friends. This type of avatar can leverage sophisticated neural network machine learning models and pre-scripted dialogue content, trained on extensive datasets, to produce unique responses.
With this type of digital avatar, users can engage in conversations, participate in relaxing activities together, share real-life experiences in augmented reality, have video calls and more. AI is crucial here, as it allows the avatars to learn about users and adapt over time.
Although the full impact of such interactions on mental health is still being researched, avatar technology — capable of tailoring responses to individual needs — can also provide significant benefits in areas such as physical therapy by capturing the actual movement of patients.
My company Robosculptor, for example, works in this space to create massage robots that are equipped with an AI-powered 3D camera, which maps a user’s body position — essentially mapping an avatar onto the synthetic image — to guide the massage session. The robotic arm monitors the user’s body posture and adjusts the treatment by responding to subtle movements.
Digital avatars can also revolutionize research and clinical trials. Using a patient’s avatar, medical workers can test the effects of different treatments to select the most effective ones. Clinicians and device manufacturers can visualize how best to deliver a treatment exactly where it is most needed. This technique is called digital human modeling (DHM). It simulates human interaction in a virtual environment.
Finally, avatars can also improve healthcare education and training. They can create medical simulations that enable healthcare professionals to practice intricate techniques in a safe and controlled setting, enhancing their skills and proficiency.
How To Create Digital Avatars
To gain insight into human behavior for avatars, it’s essential to accurately capture their form and motion. The “gold standard” for this purpose is marker-based motion capture (mocap). This process involves transforming a raw, sparse 3D point cloud into usable data. Initially, the data is cleaned and labeled by assigning 3D points to specific marker locations on the human body.
A significant challenge in capturing extensive mocap data lies in the labeling process, which, despite utilizing the best commercial solutions, often necessitates manual intervention. Issues such as occluded markers and noise can complicate matters, especially when employing new marker sets or when humans interact with objects.
Similarly, facial capture data is vital for constructing realistic human models. Unlike mocap, dense 3D face geometry is typically captured for this process. To be of use, such as in machine learning applications, raw 3D face scans must undergo alignment with a template mesh in a process known as registration. Traditional registration methods can be both slow and imperfect.
Capturing data is just the first step in creating virtual humans. Modeling involves transforming the captured data into a parametric model that can be manipulated, sampled and animated. These efforts focus on understanding human shape and movement, including its variation with different poses.
Once modeled, the avatar needs to be textured and shaded to achieve a lifelike appearance. This entails applying textures to simulate materials like skin, hair and clothing. Additionally, the avatar needs to be animated and integrated into applications or robotics solutions.
Conclusion
That said, cost is still a major concern when it comes to creating lifelike digital avatars. Skilled artists are essential, alongside a team of developers and animators. In healthcare, this concern is complicated by the need for expensive healthcare professionals, whose medical expertise is crucial. Estimating these expenses can be tricky as they depend on various factors such as project goals and industry standards.
The good news is that AI advancements offer promising solutions to cut down on development costs. Tools like ChatGPT, Midjourney and Copilot act as virtual assistants, guiding developers and speeding up tasks, thereby reducing expenses.
