Журнал дистанционного зондирования и ГИС

Абстрактный

RL-based robots for seismic sensor deployment

Ghouti L

The integration of reinforcement learning (RL) and robotics has been successfully applied in various industrial settings. One of these settings involves the deployment of seismic sensors over wide oil and gas fields. The sensor deployment problem can be formulated as a challenging optimization problem where Markov decision processes (MDPs) can be efficiently used. Our RL-based robot can deploy seismic sensors over soft and rough areas covering wide oil/gas fields. Our prototype robot resulted from an innovation work that is currently protected under two published US patents [1, 2]. A demonstration of the robot capabilities can be found in [3, 4].