- Tesla CEO Elon Musk revealed in August its plans to build a humanoid robot.
- Using Musk’s Tesla Bot, he announces that Tesla is working on exciting products for years into the future to enthuse employees, customers, and investors.
Elon Musk’s recent announcement of an upcoming Tesla Bot – an artificial intelligence robot with a human-like form and hands and characteristically optimistic delivery date – had a fair amount of criticism for a good reason.
The robot, according to Musk, will eventually be able to run errands alone, including going to the grocery store. Boston Dynamics, the company responsible for developing the most advanced robot ever made, has been working on a platform called Atlas for more than a decade. The company acknowledges that, despite impressive progress, Atlas still has a long way to go before it is physically capable of performing complex tasks on its own.
As far back as 2010, a study published in PLOS Biology offers one of the most compelling examples of evolutionary robotics potential as well as an unfulfilled promise. Researchers simulated several evolutionary models and fitness goals (not just simulations) on physical robots that contained motors and sensors: collision-free navigation, homing, predatory, and prey coevolution, among others.
They conclude that “these examples of experimented evolution with robots confirm the efficacy of mutation, recombination, and natural selection.”. As a result of their random genomes, the robots initially displayed uncoordinated behavior.”
According to the study, a few hundred generations of random mutations and selective reproduction provide sufficient time for the evolution of successful behavior in a wide range of environments.
Alphabet released more than 100 prototypes Everyday Robots, which are still very much in the prototype stage, to perform cleaning duties around Google’s offices recently. With their awkward and halting movements, the machines are still very much in the prototype stage.
The importance of progress over perfection
Unless Musk gets some help from the robots themselves, there’s a possibility that he can leapfrog the competition in the field of robotics. Experts in evolutionary computation believe that humans cannot design intelligent robots that can perform complex tasks that require constant feedback or learning loops. Robot development and design in the future may result from “evolution,” in which robots will select the most beneficial features.
It sounds like science fiction, but evolutionary robotics has existed for a long time. A century ago, Alan Turing famously hypothesized that building intelligent machines would be too complex for humans to master. The best approach would be to introduce “mutations” and selective reproduction into the designing process. Although the path to evolutionary robotics had been taking shape for quite some time, it was only recently that the tools to make it work were made available.
The foundations for evolutionary robotics have never been so plentiful: 3D printing for rapid prototyping and physical reproduction, neural networks that can be trained and used to learn, longer battery life, and cheaper materials.
Satellite antennas, for example, were developed using artificial evolution by NASA. Xenobiotics are tiny biological machines that were first simulated using evolutionary robotics techniques at the University of Vermont and Tufts University in 2020.
They demonstrated the ability to move and push payloads, suggesting that these self-healing nanorobots could deliver drugs once injected into the body.
Even with all these advancements, iterating physically is still a lengthy process, partly due to the high risks involved. In the world of robotics, even seemingly simple tasks could pose a danger to humans, like crossing a street in front of oncoming traffic.
The possibilities are endless.
It is correct that Musk describes Tesla cars as simply robots on wheels, but it is oversimplifying the situation. Despite Tesla’s specialization, they cannot self-learn how to negotiate a complex world without supervision in the absence of direct guidance. A humanoid robot capable of independently venturing into public is still probably a long way off, even though he has a supercomputer, already advanced robots, and a team of phenomenal AI experts at his disposal.
If robots perform mutations and combine desirable characteristics from different parents, one could probably create a robot that can operate independently after several hundred “generations” of evolution.
Daydream about possible real-world application areas such as surveillance and reconnaissance, building inspections and code compliance, firefighting assistance, or even search and rescue.
Nearly 100 people died in a beachfront condo collapse in Surfside, Florida, in June 2021. Building and code inspections would be an excellent example of drone swarm usage: drones could be used to perform much more frequent and regular inspections of aging condo buildings – from the top floor to the bottom, inside and out – using cameras and sensors to check for waterproofing issues, concrete spalling and cracks, sinking, and other problems. Human engineers would cost a fraction of what can be achieved with a team of robots.
Additionally, security and medical assistance are useful applications for events. Just think about the tragedy that occurred at Houston’s Astroworld. With human security personnel, covering expansive and crowded terrain at a 100,000-person event is often challenging. A robot swarm or drone can be very helpful in this regard, monitoring for security issues, fights, patients experiencing seizures, and even bringing medical equipment, such as an AED, much faster than a human staff member.
What makes drone swarms different from single drones? There are plenty of reasons, but resilience and redundancy are the main ones. The drone operation can continue uninterrupted if one fails. When the “mission” cannot be abandoned, this can be especially helpful in high-risk situations.
Making more efficient robots
A more accurate phrase would be “evolutionary robotics,” emphasizing the transfer of organic evolution to non-organic devices. In this case, the robotic development of embodied change would be a better description. The robots aren’t evolving so much as the processes are, which are growing.
A neural network and evolution algorithms could be used to create offspring from two or more parents using both mutation and subsequent recombination. Even in the absence of a physical form, evolution can solve complex problems via supercomputers. Which kinds of advancements could we achieve if we understood development better?
A real-world interaction system that is autonomous. For robots to interact with a complex, independent, real-world environment, they must be designed using evolutionary robotics. Such robots possess several advantages too lengthy to list, but some of their applications include robotic firefighters, search-and-rescue robots, nuclear waste cleanup robots, home care robots, and more.
As a result, we would also be able to understand organic evolution better. There are so many implications of an expanded understanding of evolution that it is difficult to fathom. In doing so, we may gain incredible insight into the best means of treating diseases and building immunity, improving our health and the health of others, and lessening our impact on the environment.
In addition, we could get a better understanding of life’s origins. Our ability to study and master artificial evolution will better comprehend all the possible forms and evolutions of life on other planets. According to scientific experts, the likelihood of life surviving elsewhere in the universe is low; however, an understanding of evolution and the use of microevolution as a model for macroevolution will undoubtedly assist us in our search for extraterrestrial life.
Two sides to the same coin
Lastly, consider exploring the solar system in greater detail. We could send unmanned missions to the moon far beyond our wildest imagination with fully autonomous, self-replicating, and evolving robots. Adaptive robots would use components existing on any planet, evolve according to the environment, and then send data or offspring back to Earth.
Although we are still decades away from a robot capable of learning, reproducing, observing its environment, and evolving, the idea of robots roaming the streets may turn you into a “Terminator” fan.
Instead, the biggest drawback to mastering truly autonomous robots is the inevitable displacement of the human workforce, which is unavoidable when robotics becomes genuinely independent. Musk believes universal basic income is the solution to this problem and that work will be completely optional in the future.
My opinion is not the same as yours. People feel valued by their work and contribute, and removing this source of self-worth and value could have far-reaching psychological effects as well as financial repercussions. Though evolutionary robotics may be a complex problem, it could be among the most significant accomplishments, and biggest challenges humanity faces.