AGV & AMR Robot In Warehouse

In a recent Robotics 24/7 webcast titled “How AI Is Improving and Expanding Automation,” industry experts Rand Voorhies, Chief Technology Officer at inVia Robotics, and Dan Rosenstein, Group Product Manager of Advanced Autonomy and Applied Robotics at Microsoft, shared valuable perspectives on the revolutionary advancements in AI and robotics. Here are key takeaways:

Artificial Intelligence (AI) and robotics have undergone a revolutionary transformation in recent years. Though the core of the technology remains the same, AI can now analyze vast quantities of data more accurately than ever before thanks to new tools and training models. The technology is becoming increasingly powerful. The nature of AI is such that the more it’s used, the smarter it gets, the more it’s used. It’s a virtuous cycle. It’s much like people learning a language. As more words, phrases, and syntax are learned, people make more accurate sentences and communicate better. The same applies to AI models, which improve every day. This leap in AI technology is reshaping many industries, including warehouse operations, making them more efficient, flexible, and reactive to changes in the market.

At its core, Artificial Intelligence is a technology that enables computers and machines to simulate human intelligence. This means learning, decision-making, and problem-solving, among other things. There are many types of AI: generative, predictive, optimizing, and vision AI. In fact, AI is a broad umbrella under which various technologies, such as machine learning (ML), neural networks, and deep learning, reside. Each of these technologies enables computers to learn from and interpret data, amplifying humans’ decision-making and problem-solving capabilities. And, while AI extends people’s cognitive abilities, robots extend people’s physical capabilities. They provide additional strength and endurance, making it possible to perform tasks that would have been impossible without them.

Ensuring transparency in automated systems is paramount for building trust. People need visibility into what the system is doing, why it is doing it, and what it’s planning to do. Take self-driving cars as an example: despite operating autonomously, they provide drivers with a clear view of their surroundings and intended routes. 

Similarly, inVia Robotics’ AI system offers visibility into task assignment, scheduling, and resource allocation in warehouses. Warehouse managers can rely on AI to make decisions, such as which orders to fulfill first or which tasks to prioritize based on the available resources. inVia builds trust in its AI decision-making by providing detailed dashboards and insights, allowing warehouse managers to understand why those decisions are being made.

Everybody’s talking about LLMs and generative AI, and for a really good reason: now, you can just talk to the machine, and the machine understands your intentions. The system takes raw data and provides actionable insights, eliminating the need for a BI or data analyst. What’s particularly fascinating is how advancements in LLMs are driving improvements in tooling across the board. Machine learning platform platforms like TensorFlow, PyTorch, and Microsoft Azure Machine Learning Toolkit are becoming increasingly powerful and user-friendly. You no longer need a PhD in machine learning to leverage these tools effectively.

Historically, two approaches have been used to enable communication between people and robots – creating an intermediate programming language or investing in specialized software that allows robots to communicate in a way similar to humans. And it always felt extremely unnatural. The beauty of LLMs and generative AI lies in their ability to act as a translator and “universal API”. This means that people can communicate in their natural way, either through written or spoken text, and convey information using various means such as cameras, thermal cameras, RGV cameras, depth cameras, and sound. That’s the key differentiator of the LLMs and generative AI – it has successfully bridged the gap between humans and robots. This allows people and robots to work together, and ultimately empowers both parties to perform at their best.

In the past, machine learning algorithms were designed to work only for specific use cases. If you tried to apply them to a slightly different use case, you had to re-engineer the entire algorithm by adjusting its parameters or even writing additional code. However, the future of AI is leaning towards building a model where the structure remains the same, and all the engineering gets done essentially once. This generality means that once the foundational AI model is built, it can be trained for a variety of tasks with just the relevant data. This allows companies to scale and expand without having to write a separate computer program for every use case – they just need to collect relevant training data and let the model optimize itself. This results in a significant decrease in development time and expenses for businesses, enhancing their competitive advantage in rapidly changing markets.

There are two main AI systems: AI with human involvement, known as supervised autonomy or co-bots, and AI without human intervention. As AI advances and human involvement decreases, the focus shifts from AI being perceived as magical and untrustworthy to being seen as responsible and reliable code. This transition is especially crucial as AI interacts directly with the physical world. The ability for AI to prefilter obviously or marginally unsafe solutions is particularly important, allowing for human review and intervention.

Companies that want to incorporate AI into their business model, but are new to AI should start by using tools like Azure Open AI and Bing Chat in their daily operations to see how AI can enhance efficiency and productivity, whether it’s through generating action reports, conducting internet research, or improving coding practices.

Once they feel comfortable with AI and recognize its potential benefits, the next step is to expand the focus to their teams and organizations. Explore areas where AI can streamline processes and handle tasks that are repetitive or well-suited for automation. As understanding grows and “aha” moments occur, discussions shift to orchestrating and coordinating AI solutions across multiple robots and human workers within larger environments.

Ready to transform your warehouse and elevate your team’s performance? Book a demo with us today.