RT @oxfordrobots: Wishing all the best to Team ORIon-UTBMan as they compete at #RoboCup2023 in France! 🤖✨ After two days of rigorous testin….

RT @RameshWilson1: After months of construction, HotMess has just been shipped to 9 sites in the Southern Hemisphere! 🇳🇦🇿🇦🇵🇪🇨🇱🇪🇨🇦🇷🇦🇺🇳🇿. MAS….

This work wouldn't be possible without the project team, including @robo_goga , @drbensherlock, . @CH_OceanRobots , @Alphillips200 .

In real and simulated experiments we outperform a rule-based baseline similar to how AUV missions are programmed in industry. As uncertainty grow, our method performs comparatively better. 8/9

We tested our novel planning approach on a small AUV in a deployment in Loch Ness, Scotland. 7/9

The data contents of each sensor node is modelled with a continuous time Markov process. The model is solved offline, meaning the AUV doesn’t need power-intensive onboard computation. 6/9

We use an AUV kinematic simulator to calculate transition and duration distributions. We measure navigation success by whether the AUV can successfully communicate with sensor nodes. This avoids the complexity of partially observable planning. 5/9

Our planner is designed to work with sensor nodes that provide time-of-flight "ping" localisation to the AUV – removing the need for expensive inertial guidance systems. 4/9

AUVs can get lost quickly without external position feedback. Acoustic communication isn’t always successful, and which nodes contain valuable data isn't always known in advance. Our method is the first to fully address these sources of uncertainty. 3/9.

Sensor nodes in the ocean collect data about the climate, industry or marine life, and Autonomous Underwater Vehicles (AUVs) can retrieve this data. 2/9

How do you collect valuable data from sensors far under the ocean surface? Our #IROS2022 paper's planner reasons over uncertain localisation 📍, navigation 🧭 and communication 📶 for underwater autonomous vehicles. 1/9.@GOALS_oxford @hawesie @pduckw