Volcanic emissions are a critical pathway in Earth’s carbon cycle. Here, we show that aerial measurements of volcanic gases using unoccupied aerial systems (UAS) transform our ability to measure and monitor plumes remotely and to constrain global volatile fluxes from volcanoes. Combining multi-scale measurements from ground-based remote sensing, long-range aerial sampling, and satellites, we present comprehensive gas fluxes—3760 ± [600, 310] tons day⁻¹ CO₂ and 5150 ± [730, 340] tons day⁻¹ SO₂—for a strong yet previously uncharacterized volcanic emitter: Manam, Papua New Guinea. The CO₂/S_T ratio of 1.07 ± 0.06 suggests a modest slab sediment contribution to the sub-arc mantle. We find that aerial strategies reduce uncertainties associated with ground-based remote sensing of SO₂ flux and enable near–real-time measurements of plume chemistry and carbon isotope composition. Our data emphasize the need to account for time averaging of temporal variability in volcanic gas emissions in global flux estimates.

This paper describes a series of proof‐of‐concept Beyond Visual Line Of Sight unmanned aerial vehicle flights which reached a range of up to 9 km and an altitude of 4,410 m Above Mean Sea Level over Volcán de Fuego in Guatemala, interacting with the volcanic plume on multiple occasions across a range of different conditions. Volcán de Fuego is an active volcano which emits gas and ash regularly, causing disruption to airlines operating from the international airport 50 km away and impacting the lives of the local population. Collection of data from within the plume develops scientists’ understanding of the composition of the volcano's output and is of use to scientists, aviation, and hazard management groups alike. This paper presents preliminary results of multiple plume interceptions with multiple aircraft, carrying a variety of sensors. A plume‐detection metric is introduced, which uses a combination of flight data and atmospheric sensor data to identify flight through a volcanic plume. Future work will develop the automation of plume tracking such that reliable scientific data sets can be gathered in a robust manner.

A team from the University of Bristol have developed a method of operating fixed wing Unmanned Aerial Vehicles (UAVs) at long-range and high-altitude over Volcán de Fuego in Guatemala for the purposes of volcanic monitoring and ash-sampling. Conventionally, the mission plans must be carefully designed prior to flight, to cope with altitude gains in excess of 3000 m, reaching 9 km from the ground control station and 4500 m above mean sea level. This means the climb route cannot be modified mid-flight. At these scales, atmospheric conditions change over the course of a flight and so a real-time trajectory planner (RTTP) is desirable, calculating a route on-board the aircraft. This paper presents an RTTP based around a genetic algorithm optimisation running on a Raspberry Pi 3 B+, the first of its kind to be flown on-board a UAV. Four flights are presented, each having calculated a new and valid trajectory on-board, from the ground control station to the summit region of Volcań de Fuego. The RTTP flights are shown to have approximately equivalent efficiency characteristics to conventionally planned missions. This technology is promising for the future of long-range UAV operations and further development is likely to see significant energy and efficiency savings.

The inspection of legacy mine workings is a difficult, time consuming, costly task, as traditional methods require multiple boreholes to be drilled to allow sensors to be placed in the voids. Discrete sampling of the void from static locations also means that full coverage of the area cannot be achieved and occluded areas and side tunnels may not be fully mapped. The aim of the Prometheus project is to develop an autonomous robotic solution that is able to inspect the mine workings from a single borehole. This paper presents the challenges of operating autonomous aerial vehicles in such an environment, as well as physically entering the void with an autonomous robot. The paper address how some of these challenges can be overcome with bespoke design and intelligent controllers. It details the design of a reconfigurable UAV that is able to be deployed through a 150 mm borehole and unfold to a tip-to-tip diameter of 780 mm, allowing it to carry a payload suitable for a full autonomous mission.

Fixed-wing unmanned aerial vehicles (UAVs) using conventional flight controllers are limited in their manoeuvrability, as agile flight manoeuvres require the exploitation of the non-linear post-stall flight regime. This paper extends previous work which used a Deep Q-Network (DQN) to generate open-loop trajectories for a perching manoeuvre. This paper proposes a closed-loop DQN controller, where live inference of the actuator actions is performed on a small unmanned aircraft during flight. DQN models are trained and evaluated in simulation before being deployed on the vehicle. The training processes uses a numerical flight dynamics model of the aircraft, combined with a baseline DQN implementation, to generate a series of trained models. Models with both fixed and varying start conditions at the start of each learning episode were evaluated, to identify which approach is best for generating robust controllers. Real-world closed-loop control is demonstrated through a series of flight tests, with varying wind conditions. The closed-loop controller was shown to have superior performance compared with open-loop mode, achieving a greater mean reward when performing the manoeuvre. Further, it is found that incorporating wind into the training process, as well as real-world effects such as noise and latency, is essential for further development.

Reinforcement learning has previously been applied to the problem of controlling a perched landing manoeuvre for a custom sweep-wing aircraft. Previous work showed that the use of domain randomisation to train with atmospheric disturbances improved the real-world performance of the controllers, leading to increased reward. This paper builds on the previous project, investigating enhancements and modifications to the learning process to further improve performance, and reduce final state error. These changes include modifying the observation by adding information about the airspeed to the standard aircraft state vector, employing further domain randomisation of the simulator, optimising the underlying RL algorithm and network structure, and changing to a continuous action space. Simulated investigations identified hyperparameter optimisation as achieving the most significant increase in reward performance. Several test cases were explored to identify the best combination of enhancements. Flight testing was performed, comparing a baseline model against some of the best performing test cases from simulation. Generally, test cases that performed better than the baseline in simulation also performed better in the real world. However, flight tests also identified limitations with the current numerical model. For some models, the chosen policy performs well in simulation yet stalls prematurely in reality, a problem known as the reality gap.