Around 3000 B.C., Egyptian water clocks used human figurines to strike the hour bells. In 1557, Giovanni Torriani, a Italo-Spanish clockmaker, created a wooden robot that could fetch Emperor’s daily bread from store, planting initial seeds of modern robot.
20th century has witnessed unprecedented progress; Robots have proved themselves in performing repetitive tasks with precision in controlled factory setup. Consumer robotics is expected to expand at a CAGR of 19% reaching a size of USD 17 billion by 2023, but their participation in real life is still an unfulfilled promise. Collaborative robots (cobots) having direct human robot interaction within a shared space is the next big thing which will happen to us. Here is my attempt to pen down the key considerations:
Connected Value: None of us is as smart as all of us
Covid-19 has fast forwarded the penetration of homecare applications such as robotic vacuum cleaners into the urban middle-class homes. Apart from some instances of multi-purpose robots, they largely carry out a single task. Large number of possible scenario combinations, unique environmental demands, and difficulty to program the task are key challenges to design all-rounder robots. For robots to achieve widespread adoption in daily routine they need to demonstrate collective value:
- Alternate to programming robot for each task, apply Learning from Demonstration (LFD) which allows skill transfer between humans and robots. End user to teach new tasks by familiar means producing systems that learn from skill executions led by humans.
- A collection of mobile home robots and cloud services for providing real time spatial map of connected devices can create a self-managed & responsive private space.
- Our increasing desire to speak to technology via voice platforms, smartphone, Echo or Google home provides centralized management platform for robot orchestration.
Trust is always earned never granted
In the year 2017, iRobot’s “Roomba” robot underwent severe scrutiny for collecting valuable data about its user’s homes. It’s sensitive, as internal maps prepared by vacuum cleaner for effective cleaning can be used for other purpose also. What if your robot detects absence of Sofa in living room, and you get a related ad on your social media pages? Remember, all humans have three lives: public, private, secret; and for a good reason.
Effective human-robot interaction largely depends on a level of trust that a user has in a robotic system. The important question is “Who deserves our trust, and in what respect? I trust my cook for the tasty and healthy food but may not for driving my car. While I might be fine with robot to vacuum the floors, will not trust it to administer me medicine. Trust is highly conditional and contextual; We need both, Trust in ability and Trust in intent. One needs to ascertain, “Can I trust robot’s decision? “How can I know that algorithms I am trusting are unbiased?”
Imagine a self-driving car while avoiding a child who has suddenly jumped in middle, hits another vehicle; People will judge this action differently if it was an action of a human?
To enable trust, the design of autonomous systems needs to provide capacity to explain decisions (Explainable AI-XAI) and to provide recourse mechanism when things go wrong. European Commission has published rules for ethical and trustworthy AI which is “ethical, sustainable, human-oriented and respectful of fundamental values and rights.”
Technology needs to express the human values
Our daily environment is complex, and to interact and collaborate with humans, Robot needs to develop the perceiving ability to build dynamic models of its surroundings. It involves categorizing objects, recognizing and locating humans and further recognizing their emotions. It is really complex and pushes every sub-field of robotics.
Imagine robot taking care of an elderly person during Covid pandemic who wants to go out for a walk. For safety reasons he shouldn’t be allowed. Should robot prevent him? Allow him? Inform someone? For robot to be a companion, it should take into account relevant human values; “Robotiquette” as prescribed are social rules for robot behavior as acceptable to humans.
The Final Word
For robots to be a trusted companion, definitely legal mechanism augmented by social structures for regulation and governance is needed; but building a better, unbiased AI is a long process, to be crossed by humans and robots working side by side, learning & adjusting for each other.
Researchers, industry, startups, regulatory and society at large need to drive Human–robot interaction (HRI) as a multidisciplinary social project with collaboration of computer science, psychology, philosophy and sociology. Eager to perform in all wakes of life – Manufacturing, Supply chain, Space, Aviation, Medical Science, Ambient Assisted Living, Agriculture, Education, Governance, Defence, etc. COBOTS are ready to fulfill their long pending promise!
The author is Palash Gupta, NASSCOM DeepTech Club Mentor and Director-Engineering with Huawei Technologies India.