2.1.1 Technological progress of drone logistics

AI-driven drone logistics technology has been widely adopted in the commercial field, particularly in e-commerce [Reference Rao, Gopi and Maione15, Reference Perera, Dawande, Janakiraman and Mookerjee38, Reference Tsolakis, Zissis, Papaefthimiou and Korfiatis49]. Its rapid rise signifies a profound transformation in logistics and distribution systems. As e-commerce continues to grow, traditional delivery methods have increasingly revealed various bottlenecks, such as long delivery times, traffic congestion-related delays and complex warehouse management challenges. In this context, the emergence of AI-driven drone logistics technology offers a revolutionary solution to these issues. For example, Amazon’s Prime Air project serves as a prominent case of AI-driven drone logistics in action. This initiative aims to provide fast, unmanned express delivery services, particularly focusing on achieving high efficiency in last-mile deliveries. Amazon plans to deliver goods to designated locations within 30 minutes of a customer placing an order [Reference Kocaoglu50, Reference Mandloi, Arya and Verma51]. The autonomous flight capabilities and efficient operations of drones make this project highly promising for saving time and reducing costs.

The role of AI systems in drone logistics is crucial. They not only optimise delivery routes based on real-time data but also analyse vast amounts of environmental information, such as weather conditions, traffic patterns and air congestion, using intelligent algorithms to predict and mitigate potential risks [Reference She and Ouyang19, Reference Rejeb, Rejeb, Simske and Treiblmaier37, Reference Li and Kunze46]. This ensures the safety and efficiency of the delivery process. For instance, if a drone encounters bad weather or a traffic jam, the AI system can automatically adjust the route to avoid delays or accidents. This real-time decision-making capability significantly enhances the flexibility and reliability of drone deliveries [Reference Lin, Lin and Chang52, Reference Benarbia and Kyamakya53]. Additionally, the AI-driven drone logistics system continually optimises its operational efficiency through big data and machine learning technologies. After completing each delivery mission, the drone system automatically records flight data, route information, air traffic conditions, and more, analysing this data to improve future delivery decisions [Reference Krishna Vaddy54]. This self-learning and improvement mechanism allows the drone system to gradually enhance its autonomy and task execution accuracy, paving the way for larger-scale commercial deployment [Reference Chen, Qi, Yin, Chen, Liu and Chen55, Reference Jin, Tian, Shi, Zhao, Duan and Wu56].

AI-driven drone logistics technology also plays a vital role in warehouse management [Reference Ramasamy, Natarajan and Sathyamoorthy57, Reference Manaviriyaphap58]. Unlike traditional logistics, AI-driven drones can seamlessly integrate with warehouse systems to automate the entire process from storage to delivery. Drones can quickly identify and retrieve required goods through the AI system, reducing manual intervention and optimising workflow. This not only enhances the operational efficiency of warehouses but also significantly lowers labor costs and error rates.

2.1.2 Urban and rural logistics needs

Traffic congestion in cities is an increasingly serious problem and has become a common phenomenon in many large cities worldwide. Each year, hundreds of millions of vehicles travel on urban roads, leading to significant resource waste and a decline in quality of life for residents [Reference Batty, Axhausen, Giannotti, Pozdnoukhov, Bazzani, Wachowicz, Ouzounis and Portugali59]. In this context, AI-driven drone logistics holds the potential to alleviate traffic pressure in cities through air transportation [Reference Van Hoang60, Reference Bennani61]. Drones can fly quickly over urban areas, bypassing ground traffic jams and reducing delivery times, particularly during peak hours and in busy neighbourhoods. Their flexibility and efficiency can significantly enhance delivery effectiveness. In addition, rural areas often suffer from inadequate coverage of traditional logistics networks. Many remote regions face material shortages and delivery delays due to insufficient infrastructure and logistics resources. AI-driven drones can bridge this gap by providing faster delivery services, making it easier for residents in these areas to access essential goods and services [Reference Suryanarayana8, Reference Perera, Dawande, Janakiraman and Mookerjee38]. For instance, AI-driven drones can swiftly deliver medical supplies, food and daily necessities to remote villages, greatly improving residents’ quality of life and enhancing emergency response capabilities. For example, in China, JD.com has launched several drone logistics trials aimed at exploring drone applications in rural and remote areas [Reference Xiao, Li, Cao and Xiao62, Reference Li, Shu and Wang63]. The JD.com’s drone delivery network spans multiple provinces, including Guizhou and Shaanxi, successfully addressing delivery challenges faced under traditional logistics models. In some remote mountainous regions, JD drones can quickly deliver supplies to villages despite harsh terrain and weather conditions, significantly improving distribution efficiency and accuracy. Additionally, the JD.com’s drones optimise flight routes and distribution strategies through real-time data analysis, enhancing resource utilisation.

2.1.3 Environmental protection and efficiency advantages in the logistics process

In addition to significantly improving time efficiency, drone logistics has the potential to reduce carbon emissions [Reference Chiang, Li, Shang and Urban64, Reference Goodchild and Toy65]. Traditional vehicle transportation relies on fossil fuels, leading to substantial greenhouse gas emissions and severe environmental impacts. In contrast, the battery-powered design of drone logistics makes it a more environmentally friendly mode of transport, offering unique advantages in promoting sustainable development. By utilising renewable energy sources, such as solar or wind power, to operate drones, reliance on fossil fuels can be further minimised, thereby reducing the carbon footprint of transportation [Reference Chiang, Li, Shang and Urban64–Reference Stolaroff, Samaras, O’Neill, Lubers, Mitchell and Ceperley66].

In both urban and rural delivery contexts, the lightweight design and high efficiency of drones result in significantly lower energy consumption compared to traditional vehicles. Drones do not require the complex structures and heavy engines associated with larger vehicles, enabling them to consume less energy when transporting goods over the same distance. Additionally, the integration of AI technology plays a crucial role in this process. The AI system can analyse transportation routes in real time, predict traffic flow, and intelligently plan flight paths based on weather conditions, helping to avoid unnecessary detours and delays. This enhances transportation efficiency and further reduces energy consumption [Reference Al Duhayyim, Albraikan, Al-Wesabi, Burbur, Alamgeer, Hilal, Hamza and Rizwanullah67].

With environmental protection at its core, AI technology not only facilitates the use of drones in logistics but also offers a new perspective on sustainable development [Reference Krishna Vaddy54]. As global awareness of environmental issues continues to grow, more companies and governments are recognising the importance of green logistics. As an emerging mode of green transportation, the potential of drone logistics lies not only in its economic benefits but also in its significant positive impact on the environment [Reference Chiang, Li, Shang and Urban64–Reference Stolaroff, Samaras, O’Neill, Lubers, Mitchell and Ceperley66].

2.2.1 The current and future developments of flying taxis

Companies such as Uber are developing flying taxis to leverage low-altitude airspace as a solution to the growing challenges of ground transportation [Reference Huang, Fang, Wu, Wang and Yang4, Reference Bauranov and Rakas33, Reference Marzouk68]. These flying taxis can move quickly through dense urban areas, significantly reducing travel time while alleviating ground traffic congestion. With the integration of AI technology, these aircraft will feature automated driving capabilities, which can greatly minimise human error, thereby enhancing passenger safety and travel efficiency [Reference Gordon and Lidberg69]. As computing power increases and sensor technology advances, autonomous driving technology has made significant progress. Current flying taxi prototypes are typically equipped with various sensors, including Laser Imaging Detection and Ranging (LiDAR), cameras and radar, allowing them to collect real-time information about their surrounding environment [Reference Marzouk68, Reference Pan and Alouini70]. Furthermore, AI algorithms process this data and utilise deep learning to achieve autonomy in navigation, obstacle detection and avoidance functions. Through high-precision positioning and dynamic environmental awareness, flying taxis can operate safely within complex urban airspace [Reference Rajendran, Srinivas and Grimshaw71].

In addition, the autonomous driving system continually improves its decision-making capabilities through extensive data training, enabling rapid responses to various emergencies [Reference Muhammad, Ullah, Lloret, Del Ser and de Albuquerque48, Reference Liu, Liu, Tang, Yu, Wang and Shi72]. For example, in the event of adverse weather conditions or equipment failure, the system can automatically select a safe flight path or perform an emergency landing. Advances in these technologies not only enhance the safety and reliability of flying taxis but also lay the foundation for future large-scale applications.

Moreover, to ensure safe and orderly air traffic flow, an AI-based airspace management system is also urgently needed [Reference Tang, Liu and Pan73, Reference Xie, Pongsakornsathien, Gardi and Sabatini74]. This system will monitor aircraft in real time, ensuring that they do not collide or experience other safety incidents during flight. The application of AI technology in airspace management can facilitate automated flight scheduling and route optimisation. By analysing real-time data, the system can predict aircraft operating statuses, automatically adjust routes and avoid potential air conflicts. Additionally, the intelligent airspace management system should be effectively integrated with ground transportation systems to achieve more comprehensive urban traffic management [Reference Shrestha, Oh and Kim75].

2.2.2 Air traffic safety and dispatch

In urban airspace with dense traffic, the real-time data processing capability of AI systems will be crucial for ensuring air traffic safety [Reference Kistan, Gardi and Sabatini76, Reference Degas, Islam, Hurter, Barua, Rahman, Poudel, Ruscio, Ahmed, Begum and Rahman77]. As an increasing number of drones and flying taxis operate over cities, effectively managing and dispatching these aircraft to avoid mid-air collisions and other accidents will be a primary challenge. The introduction of AI technology makes addressing this problem more feasible and efficient.

By integrating real-time data from various aircraft, AI systems can monitor flight statuses in the airspace and collect key parameters such as position, speed, heading, and altitude. This information can be used to dynamically adjust the flight paths of each aircraft, ensuring they maintain a safe distance from one another, thereby effectively reducing the risk of collision [Reference Spandonidis, Giannopoulos, Petsa, Eleftheridis and Sedikos78]. Specifically, AI algorithms can analyse historical data and real-time information using machine learning and deep learning technologies to predict aircraft movement [Reference Dong, Tao, Zhang, Lin and Ai79]. This predictive capability enables the system to make proactive decisions, such as adjusting an aircraft’s altitude or speed when encountering nearby aircraft. Moreover, the AI system can optimise aircraft dispatch based on environmental data (such as weather and wind speed) and traffic flow information, ensuring that flight efficiency and safety are given equal priority.

The application of AI technology extends beyond aircraft dispatch; it also plays a vital role in early warning and handling of potential accidents. Through advanced sensors and monitoring systems, AI can collect real-time information about the surrounding environment, including the location, speed and flight paths of other aircraft. When the AI system detects that the paths of two or more aircraft may intersect, it will immediately issue an alert and automatically adjust the routes of one or more aircraft to avoid a collision [Reference Li, Lin and Jin80]. Additionally, AI can utilise data analysis and simulation technology to construct ‘safe flight’ scenarios. By analysing flight data from various aircraft under different environmental conditions, the AI system can identify the most dangerous flight modes and conditions and formulate response strategies in advance. This data-driven intelligent decision-making can significantly reduce the incidence of air accidents [Reference Chai, Tsourdos, Savvaris, Chai, Xia and Chen81]. To further enhance air traffic safety, AI systems can also facilitate real-time collaboration and information sharing among different aircraft. When an aircraft receives scheduling information from an AI system, it not only adjusts its route accordingly but also relays its flight status and plan updates back to the system [Reference McCormack, Hutchings, Gifford, Yee-King, Llano and D’inverno82, Reference Rosenow, Lee and Li83]. This two-way exchange of information will greatly improve the efficiency of airspace utilisation and ensure that all aircraft can fly safely in dynamically changing environments.

3.1.1 Drone surveillance and privacy risks

AI-driven drones are capable of efficient wide-area monitoring and are equipped with various sensors, such as high-resolution cameras, infrared imagers and audio acquisition devices, enabling real-time observation of personal activities [Reference Cheng, Wu, Wang, Yin, Li, Chen and Chen87]. While this powerful monitoring capability enhances security, it has also raised widespread concerns about personal privacy [Reference Tan, Zhang, Li and Yu88]. The wide-area monitoring capability of AI-driven drones creates numerous application scenarios, ranging from public safety and traffic monitoring to environmental protection. However, this diversity of monitoring capabilities also introduces potential risks of privacy infringement. Drones can monitor not only public spaces but, in some cases, can illegally enter private areas to collect data, such as in residential neighbourhoods and corporate facilities [Reference Sindiramutty, Jhanjhi, Tan, Yun, Manchuri, Ashraf, Murugesan, Tee and Hussain89, Reference Geldenhuys90]. This intrusion into private spaces may lead to the leakage of personal privacy information. The risks of privacy infringement are primarily manifested in the following aspects:

• • Illegal storage and dissemination of image data: If the vast amount of image data collected by drones during monitoring is not properly managed, it may be illegally stored or disseminated [Reference Jain, Gambhir and Pandey26, Reference Ajakwe, Kim and Lee91]. If such data falls into the hands of criminals, it could result in the leakage of personal identity information, triggering a series of legal and ethical issues.

• • Malicious use of data: The data collected by drones may be exploited for malicious purposes, such as tracking specific individuals, harassment, or commercial exploitation. In these cases, the misuse of data not only violates an individual’s right to know and choose but may also cause psychological harm [Reference Yahuza, Idris, Ahmedy, Wahab, Nandy, Noor and Bala25, Reference Benkraouda, Barka and Shuaib92].

• • Expansion of monitoring scope: As technology advances, the scope of drone monitoring continues to expand. This expansion is not limited to urban public areas but may also encompass rural and remote regions. This pervasive monitoring implies that individuals could be under constant surveillance, further eroding personal privacy [Reference Schlag93, Reference Vacca and Onishi94].

• • Data security issues: During the process of collecting and transmitting data, drones face risks of cyberattacks and data leaks. If a drone system is compromised by hackers, sensitive data may be stolen, resulting in broader privacy violations [Reference Yang, Wang, Yin, Wang and Hu95]. See Table 1 for the Privacy issues and their potential consequences of AI-driven drones.

Table 1.

Privacy issues and their potential consequences of AI-driven drones

3.1.2 Transparency of data collection and use

When AI systems operate, they typically collect vast amounts of data, including video, audio and sensor information [Reference Verma, Agrawal, Patel and Patel96]. While this data plays a crucial role in the functioning of AI algorithms, it also raises significant concerns regarding the transparency of data processing [Reference Daneshjou, Smith, Sun, Rotemberg and Zou97, Reference Felzmann, Villaronga, Lutz and Tamò-Larrieux98]. Due to the complexity of the decision-making processes and the ‘black box’ nature of the algorithms, it is often challenging for the public to understand how this data is used or whether it might be shared or sold unlawfully. The opacity of data collection and processing can be reflected in the following aspects:

• • The black box effect of AI algorithms: AI systems often possess intricate internal mechanisms that make their decision-making processes difficult to comprehend from the outside. This ‘black box’ characteristic hampers ordinary users’ ability to understand how the algorithm arrives at specific decisions, especially concerning personal privacy and data usage, thereby exacerbating public skepticism and unease [Reference Yu and Alì99, Reference Rai100].

• • Lack of transparency: Most AI systems do not provide adequate transparency in their data collection and processing procedures. Users often struggle to ascertain how their data is utilised, including details about data storage, access rights, intended purposes and potential third-party sharing [Reference Daneshjou, Smith, Sun, Rotemberg and Zou97, Reference Wischmeyer101]. This lack of clarity fosters doubts about the legality and compliance of data usage.

• • Decision bias in AI systems: Since AI algorithms are typically trained on historical data, they may exhibit bias in certain instances. This algorithmic bias and lack of transparency can undermine the fairness of data processing and negatively impact specific groups [Reference Nadeem, Marjanovic and Abedin102, Reference Ntoutsi, Fafalios, Gadiraju, Iosifidis, Nejdl, Vidal, Ruggieri, Turini, Papadopoulos and Krasanakis103]. For example, in drone monitoring, algorithms may disproportionately focus on certain communities or populations while neglecting others, which not only deepens social inequality but also raises ethical concerns regarding data usage [Reference Corcoran, Denman and Hamilton104]. In particular, we can infer that without transparency, these biases can remain hidden, leading to unfair or discriminatory outcomes in critical applications such as security surveillance, logistics and urban air mobility. For example, in logistics and delivery, biased decision-making could result in inefficient routing that favors certain areas over others, potentially neglecting underserved regions. The lack of transparency also hampers accountability, as it becomes difficult to trace the origins of biased decisions or to implement corrective measures.

See Table 2 for transparency of data collection and their potential consequences of AI-driven drones.

Table 2.

Transparency of data collection and their potential consequences of AI-driven drones

3.2.1 Aircraft loss of control and public safety

Although AI-driven drones are highly intelligent and automated during flight, they can pose a serious threat to public safety if they lose control, especially in densely populated urban areas or near critical facilities. Such a loss of control can lead to property damage and directly jeopardise personal safety, potentially resulting in catastrophic consequences in the worst-case scenario [Reference Yaacoub, Noura, Salman and Chehab105]. Therefore, ensuring the safe operation of drones – particularly in public spaces – is a crucial issue that demands attention in the current technological landscape. The loss of control of drones can lead to various catastrophic outcomes, including the following:

• • Casualties: In busy city centres or at public events, an uncontrolled drone may collide with people, resulting in serious injuries or fatalities [Reference Rogers106].

• • Property loss: A drone that loses control may directly strike buildings, vehicles or other properties, causing significant damage. Additionally, it may result in secondary damage to the goods or equipment it carries [Reference Perritt107].

• • Destruction of public facilities: Near critical infrastructure – such as airports, power plants or military installations – uncontrolled drones can pose a direct threat to these sensitive areas, leading to damage to facilities or disruptions in operations [Reference Clarke and Moses108]. This not only affects the normal functioning of these facilities but may also create broader safety hazards.

• • Public anxiety and panic: Incidents involving out-of-control drones can lead to public fear and anxiety, decreasing acceptance of drone technology and potentially hindering the overall development of the drone industry [Reference Hijazi, Ferguson, Richard Ferraro, Hall, Hovee and Wilcox109, Reference Capizzi, Napoli, Russo and Wozniak110]. See Table 3 for security risks and their potential consequences of AI-driven drones.

Table 3.

Security risks and their potential consequences of AI-driven drones

3.2.2 Technical failure and responsibility

Accidents caused by technical failures during drone flights are increasingly prevalent issues, particularly within AI-driven drone systems. While AI-enabled aircraft boast advanced intelligence and the capability to operate autonomously in complex environments, they remain vulnerable to failures such as loss of GPS signals, sensor malfunctions, algorithmic errors and battery depletion. These AI-specific failures can result in loss of control, crashes or other safety incidents, posing substantial threats to personal safety and property. The integration of AI systems adds complexity to the attribution of responsibility for these accidents. AI-driven drones rely on machine learning algorithms and autonomous decision-making processes, which can obscure the direct causes of failures and complicate the identification of accountable parties [Reference Königs86, Reference Santoni de Sio and Mecacci111]. The following common situations warrant further clarification regarding responsibility:

• • Accidents caused by technical failures: If the cause of an accident is a defect in hardware design or a quality issue, the manufacturer should bear the corresponding legal responsibility [Reference Susini112, Reference Plioutsias, Karanikas and Chatzimihailidou113]. This responsibility may include recalling the product, providing compensation or implementing other corrective measures. For instance, if a battery underperforms or a sensor malfunctions, it may be deemed the manufacturer’s fault. Conversely, if the accident results from software algorithm defects – such as incorrect flight path planning or environmental misjudgement – the software developer may be held accountable. In such cases, it is crucial to review the software testing processes, update maintenance and how user feedback is handled.

• • Improper operation by the user: If a drone operator fails to adhere to safe operating procedures and an accident ensues, the responsibility should fall on the user [Reference Sabino, Almeida, de Moraes, da Silva, Guerra, Malcher, Passos and Passos114]. For example, if the operator neglects to conduct regular equipment checks, ignores weather conditions or fails to obtain necessary flight permits, these actions may result in the operator being held responsible for the incident.

• • Accidents caused by uncontrollable external factors: In certain cases, accidents may occur due to unforeseen external factors, such as sudden severe weather [Reference Oncu and Yildiz115, Reference Han, Yang, Zhao, Guan and Wang116]. In these instances, defining responsibility can be challenging. It may involve the insurance company’s responsibilities or necessitate collective reflection on the incident to promote the establishment of industry standards. See Table 4 for causes of accident evoked by AI-driven drones and the responsible party.

3.3.1 Ethical issues of autonomous combat by AI-driven drones

AI-driven drones are already widely used in military operations, including reconnaissance, surveillance and attack missions [Reference Rashid, Kausik, Al Hassan Sunny and Bappy117, Reference Abaimov and Martellini118]. Since the United States and other countries integrated AI technology into unmanned combat systems, drone tactics have significantly transformed modern warfare. With advancements in technology, the level of automation in military drones continues to increase. In the future, it is possible that manual control may no longer be necessary for conducting combat tasks autonomously. The rise of ‘autonomous combat’ drones has sparked profound questions regarding ‘war ethics’, particularly at ethical, legal, and strategic levels [Reference Mukerji119]. These drones possess the capability to identify targets, make decisions and execute attacks without human intervention, raising concerns about the extent of human control and the risk of fratricide. Without effective human supervision during the target recognition and attack decision-making processes of autonomous drones, civilians or non-combatants may be inadvertently harmed. Such incidents could result in irreparable loss of life, as well as serious international legal and ethical repercussions, damaging the credibility of the combatants involved.

Table 4.

Causes of accident evoked by AI-driven drones and the responsible party

The ethics of warfare demand that harm to civilians and property damage be minimised during military operations [Reference Epps120, Reference Boyle121]. As technology advances, the use of highly automated weapons systems may blur ethical boundaries. The deployment of drones in combat could lead commanders and operators to lose a certain level of control over the direct consequences of their actions, potentially diminishing their careful consideration of the use of force. This lack of human oversight could result in a more frequent application of force in conflicts, normalising warfare and undermining international peace [Reference Boyle121].

3.3.2 Military misuse of civilian technology

Although the AI-driven drone technology is primarily utilised for civilian purposes, its increasing accessibility has lowered the barriers for criminals to acquire or modify drones [Reference Schlag93, Reference Weiss122]. While this trend has made drones more commonplace across various fields, it has also facilitated terrorism and other illegal activities, potentially enabling the use of drones for large-scale attacks [Reference Chen, Meng, Liu, Guo and Mao123]. This not only poses a significant challenge to public safety but also places higher demands on the international community’s anti-terrorism and security management efforts. With the rapid advancement of drone technology, both consumers and companies can easily obtain various types of drones. This accessibility has contributed to the widespread popularity of drones, making it simpler for criminals to acquire, modify, or employ them for unlawful purposes.