Physical artificial intelligence (AI) is the process by which humanity's long-standing desire for someone else to do what is undesirable, tedious, or dangerous meets data, AI, and robotics and is realized in the physical world. The field where this "desire fulfillment" will first open as a market is not humanoids that replace people, but wearable robots that attach to the human body to help with walking, endurance, and recovery. Humanoids must overcome social resistance tied to labor replacement, but wearable robots start from the urgent need of patients, older adults, and industrial workers who say they "want to use my body again."
Chief Executive Cho Nam-min of Angel Robotics met with ChosunBiz at the company's headquarters in Gwangjin District, Seoul, on the 14th of last month and described physical AI this way, which has emerged as a hot topic in the global tech industry since early this year. While Generative AI, represented by ChatGPT, created text, code, and images on a screen, physical AI is the stage in which AI comes into the real world to replace or assist people.
Angel Robotics was founded in Feb. 2017 and is a wearable robot company that listed on the KOSDAQ market in Mar. 2024 under the special technology growth corporations program. It generates revenue through products including the Angel Legs M20, a robot for gait rehabilitation training; Angel Suit H10, aimed at assisting the gait of recovery and maintenance-stage patients; Angel X, a wearable suit that reduces lower-back strain for industrial workers; and Angel Kit, which bundles core robot components and modules for research and education. It has proven its technology by obtaining domestic and overseas medical device certifications centered on the Angel Legs M20 and Angel Suit H10. In Mar., it was selected for the Financial Times (FT) "High-Growth Companies Asia-Pacific 500."
◇ The wearable robot market will open before humanoids
Cho said that to understand physical AI, one must look at why robots are drawing attention "now." He said, "The term robot has been around for a long time, but it was mainly regarded as equipment that automates repetitive movements," adding, "What makes today's robots perceived differently from the past is data." As text, voice, and video have been digitized, they have become data that AI can learn, and now we have entered a stage where even human movement and interactions with the physical world are processed as data.
Cho sees that among various fields of physical AI, the wearable robot market—Angel Robotics' main focus—could open before humanoids. If humanoids enter factories or logistics sites, expectations for productivity gains will inevitably be accompanied by stronger resistance to job replacement. In contrast, wearable robots are not a technology that pushes people out; they are a technology that restores, maintains, and expands human capability, making them relatively more acceptable.
He said, "Robots should be divided into robots that replace people and robots that work with people," adding, "The technologies required for the two are different." He continued, "When a robot is worn directly by a person, even slight weight or discomfort leads to immediate rejection," adding, "The standard for the technology shifts from the machine to the human body." Whereas conventional robots focused on "position control," moving to predetermined locations, the core for wearable robots is "force control," which reads the wearer's intention and provides the necessary force.
Cho said that data is key to realizing human-centered robots. Just as language AI advanced by learning text, physical AI must learn not only human movement and intention but also functional decline and recovery processes. Cho described this as "going beyond sim-to-real (applying techniques learned in virtual environments to reality) to real-to-real (using actual usage data to further improve real-world movement)." In other words, rather than stopping at moving robots learned in simulation into reality, data generated when real people wear and move with wearable robots becomes training material for AI again.
◇ The bottleneck for spreading wearable robots is "institutions"
Cho sees that the bottleneck to the spread of wearable robots is that institutions are not keeping pace with technological progress. He said, "For medical wearable robots to become social infrastructure that reduces aging-related and caregiving burdens, rather than remaining some equipment confined within hospitals, the current medical insurance fee schedule centered on treatment acts has limits," adding, "Effects such as improved gait function, reduced risk of falls, eased caregiving burdens, and a return to daily life after discharge are hard to evaluate with the simple price of a one-time treatment act, and that needs to be reflected in the system."
Cho, who joined the company as CEO in Oct. 2024, faces the task of turning to profit. Angel Robotics posted about 4.64 billion won in sales last year and about 10.28 billion won in operating losses. While it is building a revenue base centered on medical wearable robots, losses are continuing due to expense burdens from research and development and product portfolio expansion.
To break through, Cho is focusing on overseas expansion. Since Cho took office, Angel Robotics has obtained medical device certifications for key products in Thailand, Vietnam, and Malaysia. Cho said, "We will first accumulate experience and data in the Asian market, then sequentially enter the U.S. and European markets." The following is a Q&A with Cho.
- Wearable robots also have diverse fields. Which field do you think will stand out first?
"Healthcare is the area where physical AI can most clearly prove its value. Patients must actually regain movement, and medical staff need safer, more efficient ways to support that recovery. Hospitals also want structures that raise treatment efficacy while managing more patients. It is a market where the need and demand for technology are already clear. Healthcare demands high levels of safety, efficacy, and trust. Entry is difficult, but technology validated in healthcare provides a strong foundation when expanding into other industries. We see healthcare not as a mere first sales outlet but as a market that sets the standards for physical AI."
- What area will be applied next after healthcare, and quickly?
"Industrial safety and defense. It can be used to reduce musculoskeletal strain on workers and lower fatigue from repetitive tasks. No matter how automated factories become, people do not disappear completely. Heavy loads and repetitive work still burden the human body. Wearable robots can raise worker safety and productivity together. In defense, it can evolve to reduce soldiers' load burdens and increase mobility and mission endurance."
- You cited regulation as a factor hindering the spread of wearables. What changes are needed?
"A separate pilot fee schedule and insurance system are needed for wearable robots. Rather than evaluating them only by the yardsticks of existing rehabilitation or exercise therapy, we need a separate medical technology classification and evaluation criteria that reflect the characteristics of AI-, sensor-, and data-based wearable robots.
In the initial stage, we can introduce a temporary demonstration and compensation system for innovative medical technologies, even within limited scopes. Medical institutions should be able to use the technology without expense burdens and accumulate clinical evidence. After that, based on real-world usage data, we should conduct outcome-focused evaluations such as improved gait function, reduced risk of falls, and eased caregiving burdens, and then expand insurance coverage step by step."
- Why did you first enter the ASEAN market?
"ASEAN's medical sector is evolving rapidly, and it is a market with similarities to, yet differences from, Korea. Angel Robotics sees Thailand, Vietnam, and Malaysia as key markets.
What matters is data. Until now, data from Koreans has been central. To move toward personalized wearable robots, we need movement data across diverse races, genders, and age groups. If we first build experience and data in the Asian market, we can develop more precise product strategies when later targeting Europe and the United States."
- Supply chain issues have come to the fore amid recent global instability.
"The entire robotics industry is experiencing similar problems. Prices of key components such as actuators and magnets are rising, and shifts in supply chains are another burden. For medical products, it is difficult to significantly raise selling prices just because costs have gone up. That is because there are acceptance levels among medical institutions and patients and institutional price structures. So what matters is how much we internalize core components and how stably we manage the supply chain. For wearable robots, actuators and controllers are particularly important. Simply buying parts and assembling them is not enough to build robots that attach to the human body."