Acquiring AI productive knowledge base: Pursuing drone-based applications

Like other developing countries, Bangladesh faces the challenge of developing a productive knowledge base, primarily in the technology stack that can drive the fourth industrial revolution (4IR) while leveraging innovation economy.

While innovating and making complex products may be a far-fetched dream, Bangladesh needs to keep acquiring superior productive knowledge so that the country's youths can participate in the emerging global labour market. In the era of 4IR, skill demands will progressively centre around the development of artificially intelligent (AI) components.

With a position of 103 among 128 countries in the Economic Complexity Index, this would be a daunting challenge for Bangladesh. It is widely acknowledged that increased investment in conventional educational activities will not lead to the acquisition of higher-level productive knowledge. Rather, learning by working in processes tied to developing complex products appears to be the best option for acquiring the essential skills. Such reality often poses a chicken & egg problem and finding an entry point to break the impasse is critical. This is why the immense potential of acquiring productive AI knowledge by harnessing remote sensing possibilities offered by unmanned aerial vehicles (UAVs) or drones can be considered. 

Since the first industrial revolution, the gap of accumulating productive knowledge among countries has been inceasing dramatically. The differential of holding this vital capability is largely attributed to the growing variations of complexities of products being produced and traded by different countries. It also often depends on the overseas engagement of the workforces, who return to their homelands with the experience of construction and development of complex products. The level of complexity of products being produced and profitably traded in the international market is also a measure of local value addition and per capita income contribution from industrial activities. In building higher-level productive knowledge to succeed in producing increasingly more complex products, Bangladesh needs to acquire complex productive knowledge while pursuing productive activities that will ensure profitable returns. The crucial challenge is to find the opportunities to harness the possibilities in the age of 4IR.

UAVs or drones look like flying toys, which often create nuisance in some airspace. But once they are fitted with appropriate sensors such as cameras or LIDARs (Light Detection and Ranging), they can turn into remotely controlled remote sensing platforms that facilitate the generation of a large volume of information. Through smart algorithms such as signal and image processing, pattern recognition, machine vision, and data analytics, very meaningful intelligence can be gathered through UAVs. Such intelligence could lead to a better understanding of the situation, extracting meaningful measurements and empowering machines to take artificially intelligent actions. UAV could be used in different sectors for varying purposes.

UAVs FOR PRECISION FARMING:  UAVs fitted with a camera, multi-spectral imaging sensors, or normalised difference vegetation index (NDVI) cameras can deliver images with meaningful information related to precision farming. By processing those images, real-time intelligence of crop fields associated with issues such as soil fertility, irrigation problems, fertiliser needs, pesticide attack or poor performance areas can be derived. Such intelligence can help with more uniform irrigation, precise dose determination, and early detection, consequentially leading to wastage reduction of farming inputs and higher yield. Research reports on European countries and India suggest that UAV-guided precision farming could lead to 15 to 25 per cent yield enhancement by enabling balanced use of fertiliser and/or pesticides.

SURVEYING AND 3D MODELS: Over the past 50 years, surveying and engineering measurement technology has made four quantum leaps: the electronic distance meter, GPS, mobile robotics and laser scanner. UAVs or drones will be the next quantum leap in technology.

MANAGING CONSTRUCTION AND RIVER EROSION: Roads, railways and river embankments in a flood-prone country such as Bangladesh are regularly damaged, and assessing those damages for repairs & maintenance are real headaches for government departments like Local Government and Engineering Department (LGED), Bangladesh Railway (BR) and Bangladesh Water Development Board (BWDB), among others. The use of UAVs in assessing the damage and estimating the cost could play a vital role in project planning and execution. Taking this further, recent developments in photogrammetry allow the use of (relatively) cheap camera-only UAVs (drones) to create 3D representations of infrastructure assets. These can then be imported into BIM (build information model, digital twin) systems to create scalable models and visualised in augmented and virtual reality environments, consequentially contributing to better project monitoring, quality assurance and compliance.

MONITORING GAS AND POWERLINE FOR PREVENTIVE MAINTENANCE: UAVs can stream high-definition infrared and video images of power infrastructures, which could later be processed to produce photogrammetry and infrastructure health products. For instance, a thermal camera may help identify overheating parts of infrastructure, or spots that require further action. Similarly, UAVs fitted with methane monitoring sensors could provide early warning about likely leakage of gas pipelines.

ACQUISITION OF AI PRODUCTION KNOWLEDGE FOR LEVERAGING 4IR: The journey of the exploitation of these and other possibilities offered by UAV-based remote sensing will lead to the acquisition of productive knowledge in the 4IR technology stack, needed for driving AI innovations. Some of the notable ones are: i) High-end diverse sensors, ii) Signal and image processing, iii) Machine or computer vision, iv) Photogrammetry in developing 3D and 2D models of the terrains, v) Big data and data analytics vi) Data and decision fusion, vii) Neural Network, and Deep learning, viii) Internet of Things (IoT), ix) Cloud and Edge computing, and x) Simulation, Visualization, and Digital twins. 

Acquiring a productive knowledge base in the 4IR technology stack has been a major challenge for Bangladesh. Unfortunately, the academic institutions and the classroom processes imparting knowledge and skills are extremely inadequate compared to the alternative means of acquiring those through experience and real-world applied R&D centric project-based learning. It appears that UAV-based projects hold potentials to exploit diverse possibilities. Such practicality opens the opportunity of engaging researchers, faculty members, students and also an industry to explore relevant technologies, assess capabilities, and improve them further so that they can succeed in UAV-based service innovations. As a result, among the participants, this mission will lead to productive knowledge development around these technologies. The availability of such competencies will empower existing firms as well as start-ups to recombine those bits and pieces of productive knowledge to create a larger variety of smarter and better products for leveraging the AI-driven fourth industrial revolution.

M Rokonuzzaman Ph.D is academic and researcher on technology, innovation and policy.

[email protected]

Sajjad Zohir Ph.D is the Executive Director of Economic Research

Group (ERG).

[email protected]