Towards AI-powered Endoscopic Capsule Robot Operations

Tarih: 12.03.2019
Yer: Boğaziçi Üniversitesi Kandilli Kampüsü BME Binası AZ-19



 

12 March, 2019 (Tuesday);   13.00 – 14.00 


Biomedical Engineering Institute, AZ-19,  
Boğaziçi Üniversitesi Kandilli Kampüsü, İstanbul

 

Mehmet Turan, PhD

Physical Intelligence Department
Max Planck Institute for Intelligent Systems, Stuttgart

Host: Prof. Dr. Can Yücesoy (Boğaziçi Univ. BME)

 


About The Seminar:

Capsule endoscopy has revolutionized gastroenterology screening and is often used in lieu of esophagogastroduodenoscopy and colonoscopy. Since uses and accuracy of currently available passive capsules are limited preventing them from becoming a true analogue to fully invasive procedures, active capsule endoscopic robots have been developed, miniaturized and tested for diagnosis, biopsy, therapeutic procedures and drug delivery in last years. However, active control, localization and mapping methods to be applied on these minimally invasive active capsule robots remain a significant challenge due to the facts that the gastrointestinal tract is a very complex physical environment with texture and geometry variations among different patients and heavy organ fluids causing drifts which can obstruct control, localization and mapping. This is further complicated by the peristaltic inner organ movements affecting the dynamics of the robot. Learning-based data driven control, localization and mapping methods have the potential to solve these complex issues by alleviating the need for significant engineering work. Such methods are capable of modelling highly non-linear system dynamics and agent-environment interactions, compensating for gradually changing effects, predicting and acting in anticipation of repeatable effects and disturbances not modeled prior to the deployment. This talk will be about AI-powered control, localization and mapping approaches for capsule robot operations developed by the speaker during his PhD. 

 

About the Speaker:

Mehmet Turan received his diploma degree from RWTH Aachen University, Germany in 2012 and his PhD degree from ETH Zurich, Switzerland in 2018. Between 2013-2014, he was a research scientist at UCLA. Currently, he is a post-doctoral researcher at Max Planck Institute for Intelligent Systems, Stuttgart. He is also affiliated with Max Planck-ETH Center for Learning Systems, the first joint research center of ETH Zurich and the Max Planck Society. His research interests include machine learning applications in medical robot tasks such as learning based SLAM (simultaneous localization and mapping) methods and deep reinforcement learning based control approaches for milliscale medical robots. He received DAAD (German Academic Exchange Service) fellowship between years 2005–2011 and Max Planck Fellowship between 2014–present. He has also received Max Planck-ETH Center for Learning Systems fellowship between 2016–present.