By: Peter Karanja, Autonomy Global Ambassador – Africa
Autonomy is easy to sell on a slide deck. I have seen potential customers light up when they learn that artificial intelligence (AI) systems can automatically detect and flag anomalies with little to no human input. In one instance, a customer said that they wanted a fully autonomous AI system to detect anomalies during the inspection of a power line because they didn’t “trust” humans to detect them accurately. While this sounds good on paper, the real wins in Africa are not coming from full autonomy. They are coming from human-in-the-loop (HITL) systems.
What Are HITL Systems?
HITL systems are AI and Machine Learning (AI/ML) systems that involve humans to ensure accuracy, improve efficiency and reduce errors. Instead of thinking of full autonomy as the end goal where the fewer humans involved, the more advanced a system is, Africa should optimize for something different: reliability, recoverability in case something goes wrong, compliance and safety and costs per mission.
Don’t get me wrong. Full autonomy works for repeatable tasks such as executing a waypoint mission, navigation, obstacle sensing, data capture and geofencing. But the human element is still useful in decision-making, mission context, determining when to grant exceptions and to ensure the system’s safety.
Why HITL Works for Africa
Africa faces unique challenges that make HITL systems more ideal. Let’s look at some of them.
Inconsistent Connectivity
To run a fully autonomous system in Africa, you need adequate connectivity for cloud processing, fleet monitoring, continuous telemetry, remote operations centers and real-time updates. For Africa’s real-world operations, connectivity can be patchy, expensive to set up, lack enough bandwidth and unavailable for long-range missions. If your system requires persistent connectivity throughout, it will struggle. In such situations, you will need systems that can be controlled manually in case connectivity is lost.
Inadequate Power and Charging Logistics
When we talk about autonomy, we are not only talking about flight control. We are also talking about charging cycles, maintenance, battery health, repairs, storage and transport. In actual field operations, power is not stable enough to ensure smooth operations without planning. What actually works is predictable mission planning that revolves around actual flight times. Secondly, operational standard operating procedures (SOPs) should be designed for accurate charging windows. And finally, the autonomy that works here reduces repeated work due to re-flys, missed captures and unstable routes.
Airspace and Compliance Considerations
In many autonomy discussions, airspace regulations are treated as downstream considerations after the technology proves itself. But in many real-world applications, regulations are part of the discussion about whether operations scale. If the system can’t operate legally, safely and confidently in shared airspace, then it can’t scale, regardless of how good the technology is. Autonomy is shaped by:
- Approvals: Before starting the operations, someone must approve the aircraft, operator, mission and location. If autonomy can’t clearly define who will be in charge of the mission and who will be responsible in case something goes wrong, it will be difficult to approve.
- Operational limits: Every operation has boundaries such as altitude, location, weather, visual line of sight (VLOS) vs beyond VLOS (BVLOS) and proximity to people or infrastructure. An autonomous system should be able to adhere to these limits.
- Documentation: For every operation, the authority will ask for SOPs, emergency procedures, risk assessments, operator qualifications, flight logs, etc. Without proper documentation, approvals, insurance and audits, execution will be difficult.
- Stakeholder confidence: Often, stakeholders want an autonomous system, but when something goes wrong, they need someone to hold accountable.
HITL systems increase confidence in regulators and other stakeholders because accountability is visible. Combining people with actual knowledge of ground operations with autonomous systems enables you to explain procedures, justify safety measures and align operations with local regulations.
Unstructured Environments
Full autonomy works best in structured environments, such as in predictable terrain, mapped obstacles, controlled roads and controlled airspace. But many African operational settings are characterized by dynamic obstacles, unpredictable weather, irregular terrain and limited mapping data for obstacles. In such uncertainty, 100% autonomy would not be practical. Instead, safe autonomy with human override would be ideal.
Inadequate Domain Expertise
Most of the autonomous systems being introduced in Africa have been trained without knowledge of Africa’s local setting and regulations. As a result, a HITL system needs to be adapted during AI training or during operations.
HITL Systems in Africa Today
Let’s look at what is currently working on the ground regarding autonomous systems:
Health Access and Logistics
Africa consists of many rural areas that are hard to reach by road and too expensive to reach by manned aircraft. Drones offer a more accessible way to deliver critical medical supplies to hard-to-reach areas. Ideally, medical delivery drones will be autonomous to ensure efficiency, minimize risk to human pilots in hazardous areas and maintain consistent flight control. However, even such autonomous systems need people to oversee them. For instance, Zipline, one of the largest medical drone delivery companies operating in Africa, uses autonomous drones, but local health workers, drone pilots and aviation professionals manage order delivery, pre-flight checks before take-off, mission control and monitoring and retrieval. The paths that these drones follow are pre-approved by the civil aviation authority (CAA), which underscores the importance of people in ensuring regulatory compliance.
Disaster Response and Search and Rescue
Autonomous systems can quickly scan disaster-stricken areas for mapping and search-and-rescue (SAR) operations. However, human input is still necessary to verify what the drone sees and ensure rescue efforts are coordinated effectively.
Agriculture and Food Security
Farmers in Africa, especially small-scale farmers, often lack enough data to optimize their yields. Drones equipped with AI/ML can collect data on these farms so agronomists, farmers,and other stakeholders can use it to make informed decisions. On the operations side, no farm is the same shape as the other. As much as you can plan a flight plan for the drone to execute automatically, a qualified pilot needs to monitor the operations and be ready to take manual control in case of an incident.
Environmental Conservation and Monitoring
Drones have been effectively used to map and identify invasive species as well as prevent poaching by providing aerial surveillance. Still, experts in animal behaviour and ecosystems play a significant role in interpreting drone data, classifying false positives and documenting incidents to support better decision-making. Experts in drone operations will be needed to adapt drone operations to the specific environment in which they will be flying.
Security and Public Safety
Both private and public stakeholders in Africa are exploring how they can use autonomous systems to improve the efficiency of their security operations. However, the use of drones in a security setting requires constant human oversight to address ethical concerns, privacy issues, and misuse. HITL systems in such cases help ensure accountability and regulatory compliance.
Infrastructure Monitoring and Maintenance
Africa has extensive infrastructure, including pipelines, wind turbines and railway lines. Inspecting this infrastructure manually can be quite difficult, especially in remote areas. Drones can be used to monitor this infrastructure and automatically identify any anomalies. Technicians can then analyze the results to verify and determine which areas need immediate maintenance.
The Real Deal
Autonomy adoption in Africa is not behind. It’s adapting to real operations. From health and logistics to security, autonomy handles the repeatable tasks while humans still must handle judgment, exceptions and accountability. To scale autonomy in this vast continent, you don’t need to chase full autonomy at all costs. You need to chase practicality. This translates into: HITL systems, resilient workflows, autonomy designed for imperfect infrastructure, adequate safety redundancies and measurable operational outcomes. Autonomy is not a magic switch that replaces humans and suddenly makes everything move smoothly. It is a set of tools built based on operational reality…especially here in Africa.