The Autonomous Age: Robotaxis, Drones and Logistics

A New Autonomous Era

For a while now, we have been transitioning from the industrial age of the human-controlled self-propelled machine (tools, boats, cars, and planes) to the self-acting machine (industrial robots, GPS-controlled airplanes or tractors, etc.).

However, there were still a lot of limitations on the capacity of machines to act without human supervision. It relied either on a very controlled enthronement like a factory floor or on a very simple set of instructions, like going from point A to point B following a rudimentary path.

This was due to a few reasons:

• Machines cannot see their environment.
• Machines are big, heavy, and powerful, making them too dangerous to not be controlled by humans out of a very narrow set of situations.
• Machines cannot think and, therefore, cannot react to unexpected circumstances, hence the need for humans.

But all of this is now changing.

Machine vision is now advanced enough that detecting obstacles is mostly a solved problem. Humanoid robots, small drones, and farming robots are much less threatening than industrial robots or tractors weighing tons. But more importantly, AI is now giving machines an unprecedented level of intelligence.

The Autonomous AI Revolution

In 2023, people have been shocked by how human-like LLMs (Large Language Model) systems like ChatGPT sound. Similarly, AI art has progressed exponentially.

There are already concerns that AIs might make entire professions like writers, teachers, or illustrators entirely obsolete.

But it is not only in the world of ideas and concepts that AI thrives. With increased levels of intelligence, AI is now able to navigate the real world of streets, warehouses, and even battlefields.

Together with progress in batteries (more autonomy), EVs (no need for combustion engines), and captors (cameras, LIDAR, and other detection methods), the increased computing power and computing efficiency make possible a multitude of concepts for now reserved to the domain of science-fiction:

• Self-driving cars, as well as self-driving trucks, self-piloting ships, and maybe even planes.
• Drones deliveries, both on wheels and in the air.
• Autonomous combat drones.
• AgTech drones.
• Etc.

Source: ARK Invest

(In this article, we will explore the different concepts by order of technological maturity, from the already in deployment to the one still in development)

Autonomous Logistics

Self-Sorting Warehouses

The first step of modern logistics is in giant warehouses, gathering hundreds of thousands of different items, as well as sorting facilities of e-commerce giants and postal services.

The job of finding, picking, and moving to the right spot the goods and parcels is increasingly handled entirely by robots. This can range from small box-picking robots, to forklift-like larger ones. Companies like Amazon (AMZN) are increasingly relying on these systems for their warehouse jobs, including fully autonomous robots developed together with Agility Robotics. So is Alibaba (BABA), or AutoStore Holdings Ltd. (AUTO.OL).

This is by far the most established use of autonomous robots, with for example Amazon developing this field for the last 10 years. Soon, Amazon might be “employing” more robots than humans.

Source: ARK Invest

Autonomous Deliveries

Another activity that is labor-intensive in logistics is last-mile deliveries. This includes e-commerce but also food delivery and local truck traffic. Depending on the item, these activities could cost from 57% less to 94% less if they were handled by autonomous systems.

Source: ARK Invest

This represents an absolutely massive potential market, which ARK Invest estimates could grow from the current almost zero to $900B in 2030. In total, the addressable market could be in the $1-2T size.

Source: ARK Invest

This is a segment where proprietary data for training the AIs is very valuable. For now, the leaders in rolling robots are Estonian Starship Technologies launched by Skype co-founders, and Chinese e-commerce giant Alibaba; both have adopted a small and harmless design not dissimilar to Star Wars' droids.

Source: Starship

Due to regulatory reasons, autonomous trucks are still relying on human input, with the company Kodiak leading the way, followed by Gatik and Pony.ai.

Another idea is to rely on flying drones for performing the delivery of small and light-enough items. So far, the leader is clearly Zipline, followed by Wing and Meituan (3690.HK). This could be a true revolution, but faces even more regulatory hurdles than self-driving delivery trucks, so might be slower to be deployed at scale.

Source: ARK Invest

In some circumstances, the flexibility and mobility of flying drones can be literally life-saving. For example Zipline deliveries of blood for transfusion in less than 15 minutes, reducing post-partum hemorrhages by 80% in Rwanda.

Source: ARK Invest

Other Self-Driving Logistics

Self-Driving Trucks

Self-driving long-haul semi-trucks are another part of the logistical chain that could be turned autonomous. For now, the progress has been limited by the slower-than-expected progress of self-driving cars, relying on the same technology. And trucks being much larger and heavier are in need of an even more certain and safe self-driving technology than cars.

However, we might see some level of self-driving trucks soon, at least on highways. The company Aurora Innovation is looking at launching the first autonomous trucks on highways in 2024.

Source: Aurora

As a more controlled environment than streets with bikes, dogs, children, and pedestrians, highways are probably the ideal environment for the launch of autonomous trucks. And represent the bulk of the kilometers driven by a human requiring to be paid by the hour and getting tired over time.

Self Driving Ships

Another manpower-intensive logistical task is manning the ships that cross the oceans and keep the global trade flowing. Rolls-Royce (RYCEY) is a company known for its luxury cars, or maybe by some people for its plane engines. However, it is looking at becoming a leader in autonomous ships, with DNV as another contender. Rolls-Royce envisions a roll-out of remote-controlled coastal ships in 2025, remote-controlled ocean-going ships by 2030, and autonomous ocean-going ships by 2035.

Source: Rolls Royce

Self-Driving Cars And Robotaxis

While carrying packages is certainly very useful, and likely to be very profitable, a large demand for transportation is focused on people.

With most households in developed economies owning one or several cars, and commuting to work, schools, and leisure, there is a massive untapped demand for self-driving cars.

The cost of personally owned vehicles has pretty much stagnated since the first Ford cars in 1934, at $0.7/mile. But autonomous taxis could bring it down to $0.25/mile. This is because a robotaxi cuts costs in two ways: first by removing the need for a driver, and second by allowing for a much higher utilization rate of the car, which can essentially run 24/7.

Currently, robotaxi rides' total miles are dominated by Chinese Baidu, followed by Google-backed Waymo. Meanwhile, another company in the sector, Cruise, has stopped US operations.

Source: ARK Invest

A strong contender to arrive in this market will be Tesla, which is collecting data from all of its existing fleets for free, differentiating itself from its competitors by using custom vehicles for acquiring data to train their AIs. In a way, Tesla customers are all working for free for the company to collect real-life traffic data.

Tesla's self-driving program is also looking to rely exclusively on cameras, mimicking the way a human does, instead of more expensive (several times the price of the car) LIDAR systems. With 50x more driving than Baidu, if one company can pull off full self-driving beyond pre-established narrow paths in specific cities, it will likely be Tesla.

Self-Driving Safety

A key element for the authorization & adoption of self-driving is safety. Due to our innate skepticism in letting a computer free to drive a heavy and dangerous car, robotaxis need to be at least an order of magnitude safer than human drivers in order to be accepted.

And this seems to be happening, with a Tesla in FSD (Full Self-Driving) 5x safer than a Tesla under manual control. It is worth noticing that this also makes Tesla's FSD much safer than Waymo's. Overall, the more progress in AIs is made, the more self-driving will improve, including interaction through natural language thanks to LLMs.

Source: ARK Invest

By reducing cost per mile, autonomous driving will likely change the calculus around car ownership and strongly boost the ride-hailing market. With the forecasted $0.25/mile, the ride-hailing total addressable market (TAM) could grow from the current $4-100B (depending on ride price) to $5T. And makes most people give up on owning a car, except if it is to “rent” it out as a robotaxi when they are not using it.

Source: ARK Invest

Autonomous And Precision Farming

Farming has increasingly become a high-tech activity, relying heavily on satellite images, heavy machinery, genetically optimized seeds, complex chemical treatments, and precision software.

This trend is likely to continue with the emergence of advanced farming robots. This is already partially implemented with limited self-driving capacity for tractors, combines, and other similar equipment, with John Deere (DE) as one of the leaders of the concept.

The next step will likely be much smaller and smaller robots. This is true in open fields for everything from weeding (possibly with lasers like with Carbon Robotics) to planting, harvesting, watering, etc. Another labor-intensive activity on the way to be automatized is fruit picking, like with the 6 robotic arms apple picking system from Advanced.farm.

You can read more about this revolution in farming in our articles “Investors Should Take Note: Robotics Is Taking Over Farming” and “Revolutionizing Agriculture: The Role of Robotics in Boosting Productivity and Sustainability”.

Autonomous Weapons

There is a deeply unsettling idea of creating a machine that can decide autonomously how to move, fire, and kill humans. Decades of science fiction stories like Terminator do not help either.

Nevertheless, military commands all over the world are looking at the option. In large part because drones and robots are much more acceptable (and cynically, often cheaper) loss than human soldiers.

And in part because they might be better at the job, with AI now outperforming human piloting capacity for example.

Source: ARK Invest

Giving AI autonomous capacities can also increase the efficiency of an army. The war in Ukraine is a good example: the Russian army has been deploying self-driving tracked robots for ammo deliveries and injured evacuation, and has also given an AI-targeting identification system for its Lancet suicide drone. Meanwhile, the Ukrainian forces used submersible drones to hit targets in the Black Sea.

So as with many innovations, it is likely that autonomous vehicles and AI progress will have a dark side, boosting the destructive power of humans against each other.

Humanoid Robots

When it comes to autonomous systems, for now, the target has been to make autonomous existing systems, like cars, drones, or delivery systems. But ultimately, the goal is to create humanoid robots that could seamlessly interact with us.

This is key to realizing the full potential of autonomous systems, as our entire world and infrastructures are designed for human bodies.

By 2030, ARK Invest expects that by 2030 the “annual unit sales of humanoid robots have grown to 10% of the number of humans in the manufacturing workforce”. And that by 2040 “The cost of humanoid robots that are backward-compatible with existing infrastructure has dropped below that of human manufacturing labor for many applications”.

And this is not as far-fetched as it sounds like. Boston Dynamics Atlas can move and jump like a human (Boston Dynamics was acquired by Hyundai – HYMTF).

Source: Boston Dynamics

And Tesla has been working on a humanoid robot as well, although with slower progress than expected.

So if delivery robots for food are already a reality, and self-driving cars are soon to be one as well, a world populated by both humans and “droids” might be within reach in less than 10-15 years.