Hidden Challenges of Deploying Robots in the Real World
Part 5: Workflow Integration
Arguably the hardest part of deploying robots in real-world settings is figuring out the practical details of how they can be used effectively within everyday workflows. Robots are not direct substitutes for people . Existing workflows often need to be adapted or redesigned, which is especially challenging in environments like hospitals, where staff have ingrained habits developed over decades.
Adding to this complexity, hospitals were not designed with robots in mind. Each site has unique physical structures, workflows, and key individuals who influence operations. As a result, ensuring that robots integrate seamlessly into daily activities requires careful thought, flexibility, and collaboration. In this installment, I’ll reflect on some of the operational challenges we’ve faced and the steps Akara has taken to ensure our solution meets the nuanced needs of our customers.
Storage and Charging
Hospitals are often space-constrained, so finding appropriate storage for robots , both during and between uses , can be surprisingly difficult.
Storage When Not in Use: At the end of the day, the robot needs a ‘home’ base where it can be securely stored and charged. For us, this meant identifying locations that weree secure and inaccessible to the public, near the areas where the robot is needed to minimize transit time and equipped with a charger that staff could easily connect to.
Storage Between Uses: During operational hours, the robot may need to wait between tasks. This is often done in spaces adjacent to the rooms where the robot is required, ensuring minimal disruption to hospital workflows.
Charging is another critical consideration. A low battery doesn’t just impact operations ; it can pose safety risks if a robot unexpectedly shuts down in a busy environment. While automatic charging docks might seem like the obvious solution, hospitals are protocol-driven environments, and introducing autonomous charging adds risk and disruption. Instead, we opted for a human-in-the-loop approach, where staff ensure the robot is charged at the end of the day, minimizing risks while keeping operations simple and reliable.
How the Robot Interacts with Staff
Deploying a robot is as much about adapting workflows as it is about the robot itself. Practical questions , such as how staff summon, interact with, or move the robot require carefully designed solutions to ensure the system works seamlessly in day-to-day operations.
At Akara, we developed multiple interaction methods to suit different workflows and preferences:
Manual Movement: Sometimes the simplest solution is best. Staff can manually push the robot into place or use a remote control to guide it. This approach is particularly useful in hospitals where robots cannot autonomously open doors or when quick deployment is needed.
App-Based Summoning: To improve efficiency, we built apps, available on web browser and iOS/Android , that allows staff to summon the robot autonomously, much like calling an Uber. The app provides a familiar interface that requires minimal training.
Hybrid Approach: In workflows requiring a combination of manual and autonomous operation, staff move the robot part of the way (e.g., disconnect it from charging) before handing over control to the robot’s AI system. This hybrid approach ensures flexibility in environments where fully autonomous operation isn’t feasible.
Fallback Behaviors
In dynamic hospital environments, robots must be able to detect issues and safely adapt their behavior when something unexpected happens. To ensure reliability and safety, we implemented fallback behaviors for two key scenarios:
Detecting People: Robots operating in environments with people need to adapt their behavior to ensure safety. For example, Akara’s robots use AI to reliably detect people nearby. When someone is detected, the robot takes precautionary steps such as:
Turning off UV light to prevent accidental exposure.
Issuing visual or audio warnings to alert staff.
Handling Unexpected Events: Robots must be capable of responding gracefully when something prevents them from executing their tasks. Common scenarios include:
Getting stuck or encountering unexpected obstacles.
Losing internet connectivity.
Malfunctioning sensors or hardware.
In such cases, the robot’s software triggers safe fallback behaviors , like stopping, issuing alerts, or returning to a default state , ensuring that failures do not escalate or disrupt operations.
IT Systems Integration
Integrating robots into hospital IT systems is often necessary to unlock their full potential, but this can be one of the most complex and time-consuming parts of deployment. There are three primary systems where integration is commonly required:
Electronic Health Records (EHR): EHR integration allows the robot to interact with patient and room status data. For example, the robot might receive a notification when a room is ready for cleaning or send a completion message back to the EHR. This involves:
Working with EHR provider APIs to extract or push data.
Navigating challenges with smaller EHR providers, whose APIs may be less mature.
Engaging with larger EHR systems like EPIC, which require vendor approval, HIPAA compliance reviews, and integration into their complex software ecosystems.
Building Management Systems (BMS): BMS integration can enable tasks such as opening electrically controlled doors or transmitting sensor data (e.g., air quality). However, many BMS systems are not designed with robots in mind, and solutions like OpenRMF , which offer guidance for robotic facilities integration, are not yet commonplace. Additionally, older hospitals often require capital investment to make systems robot-friendly.
Third-Party Device Infrastructure: Integrating with third-party devices (like sensors or access control systems) adds another layer of complexity, as these systems often have their own protocols and hardware requirements.
By prioritizing versatile integration frameworks, we’ve ensured our robots can adapt to varying IT environments, even when systems are fragmented or outdated.
Conclusion
Looking back on this journey, it’s clear that building hospital robots isn’t about flashy tech. It’s about understanding and adapting to the complexity of the social, physical and dynamic environments they operate in. Every challenge we’ve faced, from navigation to integration, has deepened our appreciation for the real-world systems and people our robots work alongside. These experiences continue to shape how we build, not just for functionality, but for trust, safety, and lasting impact, all necessary ingredients for lasting adoption.