Stochastic Guidance of Buoyancy Controlled Vehicles under Ice Shelves using Ocean Currents

TL;DR

The paper addresses autonomous guidance of buoyancy-controlled under-ice vehicles to reach grounding zones for in-situ melt-rate measurements amid uncertain currents. It introduces a model-based framework using approximate dynamic programming to synthesize depth-control policies from a stochastic under-ice flow model, with a QMDP extension to handle localization uncertainty, validated on MITgcm Pine Island cavity simulations. The results show up to $88.8\%$ of vehicles reaching the grounding zone (a $+33\%$ improvement over state-of-the-art, $+262\%$ over uncontrolled drifters), and performance improves toward $95\%$ as localization improves, demonstrating practical potential for cost-effective, scalable under-ice data collection. The approach enables reliable access to ice shelf grounding zones, improving melt-rate measurements that inform sea-level rise projections and climate science.

Abstract

We propose a novel technique for guidance of buoyancy-controlled vehicles in uncertain under-ice ocean flows. In-situ melt rate measurements collected at the grounding zone of Antarctic ice shelves, where the ice shelf meets the underlying bedrock, are essential to constrain models of future sea level rise. Buoyancy-controlled vehicles, which control their vertical position in the water column through internal actuation but have no means of horizontal propulsion, offer an affordable and reliable platform for such in-situ data collection. However, reaching the grounding zone requires vehicles to traverse tens of kilometers under the ice shelf, with approximate position knowledge and no means of communication, in highly variable and uncertain ocean currents. To address this challenge, we propose a partially observable MDP approach that exploits model-based knowledge of the under-ice currents and, critically, of their uncertainty, to synthesize effective guidance policies. The approach uses approximate dynamic programming to model uncertainty in the currents, and QMDP to address localization uncertainty. Numerical experiments show that the policy can deliver up to 88.8% of underwater vehicles to the grounding zone -- a 33% improvement compared to state-of-the-art guidance techniques, and a 262% improvement over uncontrolled drifters. Collectively, these results show that model-based under-ice guidance is a highly promising technique for exploration of under-ice cavities, and has the potential to enable cost-effective and scalable access to these challenging and rarely observed environments.

Figures (5)

• Figure 1: IceNode concept of operations and mission phases
• Figure 2: Instantaneous currents under the Pine Island ice shelf at $500$ m depth and across one vertical slice for one time step.
• Figure 3: MDP problem state value for $z = - 500$ m. The color of each location denotes the expected discounted reward obtained when following the optimal policy from that location.
• Figure 4: MDP problem policy for $z = - 500$ m. Color denotes the change in depth prescribed by the optimal policy.
• Figure 5: Policy rollouts and time required to reach the landing zone for successful rollouts. For each policy, 500 IceNode trajectories are simulated. The color of the trajectory shows the change in depth, either through a control action or vertical forcing due to current: yellow corresponds to an ascent, blue captures constant-depth drifting, and cyan shows a descent. Red dots show the vehicles' final locations.