Suggestions
Joanna Lichter
Venture Fellow at Pear Vc
Highlights
If the skilled labor shortage remains unchecked, there will be millions of unfilled jobs in construction + manufacturing by 2030. As highlighted previously, there is a growing mismatch between skills required on the job site + those available in the workforce.
No VC post would be complete without mentioning AI , + it’s not hard to imagine how this could impact labor + productivity –particularly in robotics, where AI-based systems are streamlining labor-intensive tasks in construction, from assembly to material transport.
Historically, robotics were largely limited to well-defined, “structured” tasks. However, advancements in AI are rapidly expanding robotic capabilities to handle “unstructured” use cases, where robots have some ability to “think” for themselves + make decisions in dynamic environments, such as construction sites.
What's driving this progress?
🧠 Generalizable, multimodal foundation models such as Google’s RT-2 and PaLM-E are enabling robots to perform complex tasks (e.g. stacking bricks) by breaking these down into simpler steps. Using LLMs as a foundation, robots can then be trained using a combination of text, images, + robotic movement primitives to learn + execute specific actions. 🌐 Simulators that leverage physics-based modeling are enabling more precise, real-world experimentation. Newer simulators can even generate datasets to train ML models that later run on physical robots.
🗺 Improvements in managing uncertainty - algorithms like SLAM enable robots to create + update maps of an unknown environment while simultaneously keeping track of their position, while SLAP layers on cost-efficient path planning –which allows robots to navigate routes while avoiding obstacles.
These developments, alongside declining hardware costs, are broadening the scope of robotics use cases in construction + beyond. I'm particularly excited by generalizable software platforms that can be applied across various robots vs. bespoke solutions that work with one type of hardware, as is standard today.
For example, companies like Field AI are building autonomy software to empower robots to thrive in unstructured environments. These robots can be used to autonomously capture on-site data + monitor building progress –typically, very tedious + expensive tasks handled by project engineers, who must manually walk sites + take photos. Others like AIM Intelligent Machines are retrofitting existing machines such as excavators with sensors, + combining these with ML algorithms to turn these into fully autonomous systems.
We're just beginning to scratch the surface of what robotics can do in construction. While the workforce shortage underscores the need for more R&D, solutions must be economically feasible, integrate within existing workflows, + gain cultural acceptance on-site. I look forward to backing the next generation of founders innovating here –please reach out if this is you!
- Thank you to Malik Ismail for your incredible research! 🙏
📣 It's that time of year again!
Emerson Collective is looking for 4 exceptional MBA students to work with our venture team this Summer. Our MBA Associate Program is a great opportunity to help us find, diligence, and support the best companies delivering generational impact and venture-scale returns.
The deadline to apply is February 12, 2024.
Apply here!
