ATLAS is looking for outstanding candidates to fill in 15 Early Stage Researcher positions.

The proposed vacancies are Joint Doctorates. This means that part of the PhD will be hosted at the Main Institution, while part will be hosted at the Secondary Institution (a period between 6 month and 1 year).

The ESR will have the opportunity to achieve a double or joint PhD degree, fulfilling the requirements of both institutions.

Each ESR project foresees one or more secondments to relevant companies or research centre, not necessary in the same country of Primary or Secondary Institutions. Candidate mobility, as well as motivation, is a requirement.


Computer vision and machine learning for tissue segmentation and localization

Each flexible robot will be equipped with extero- and proprioceptive sensors (such as FBG) in order to have information on position and orientation, as well as on-board miniaturized cameras. Additionally, RT image acquisition will be performed using US sensors externally placed in contact with the patient outer body. In order to track the position of the flexible robot and to simultaneously identify the environment conditions RT US image algorithms will be developed. Deep learning approaches combining Convolutional Neural Networks (CNNs) and automatic classification methods (e.g. SVM) will extract characteristic features from the images to automatically detect the:

  1. flexible robot shape
  2. the hollow lumen edges positions (to be integrated with ESR5) and
  3. information on surrounding soft tissues shape and location.

RT performance will be achieved by parallel optimization loops.

Main  institution and supervisor: POLIMI, E. De Momi

Secondary institution and supervisor: UNISTRA, M. de Mathelin


Simultaneous tissue identification and mapping for autonomous guidance

One of the challenges of autonomous surgical devices is the ability to navigate towards the clinical target. Two aspects have to be addressed to answer that problem.

First of all, targeted disease has to be identified in the organ of interest. Second of all, because the large majority of luminal organs inside of the human body have a very complex geometry, real-time information about position of the device in the lumen is needed. In this project, a side-viewing OCT catheter will be used in conjunction with a robotic endoscope to collect images, which can be used for simultaneous assessment of the clinical status of tissue and recovering of information about the position of the instrument in the lumen.

The position mapping can be further improved by combining optical information with robotic measurements. Additional imaging and sensing modalities like electromagnetic tracking or IVUS may also be used.

Main  institution and supervisor:UNISTRA, M. de Mathelin

Secondary institution and supervisor: UNIVR, P. Fiorini


Image-based tool tissue interaction estimation

One of the keys to bring situational awareness to surgical robotics is the automatic recognition of the surgical workflow within the operating room. Indeed, human-machine collaboration requires the understanding of the activities taking place both outside the patient and inside the patient. In this project, we will focus on the modelling and recognition of tool -tissue interactions during surgery.

Based on a database of videos from one type of surgery, we will develop a statistical model to represent the actions performed by the endoscopic tools on the anatomy. We will link this model both to formal procedural knowledge describing the surgery (e.g. an ontology) and to digital signals (such as the endoscopic video) in order to provide information that is human- understandable. Models and measurements coming from the robotic system will also be incorporated, in order to develop semi -supervised on unsupervised methods, thus limiting the need to annotate the endoscopic videos.

Applicants should have a MSc/MEng (or equivalent) in Engineering, Computer Science, Mathematics or related disciplines. Applicants must have strong programming skills and background in Computer Vision and Machine Learning.

Main  institution and supervisor: UNISTRA, N. Padoy

Secondary institution and supervisor: POLIMI, G. Ferrigno


  1. 55082145
  2. 55013198
  3. 55113400
  4. 54786629


Surgical episode segmentation from multi-modal data

In order to decide which actions to perform, an autonomous robot must be able to reliably recognize the current surgical state or phase it is in . This is especially true in a context of shared autonomy, where part of the procedure is still done by a human operator, or if the robot is to intelligently/semi-automatically assist manual gestures performed by a surgeon. To this end, deep learning methods (e.g. combined CNN for video and RNN for lower dimensionality data) will be applied for extracting the relevant information. The originality of this ESR project is that multi-modal data will be considered: building from previous developments on phase detection in endoscopic video data, the algorithms will be augmented with data coming from intra -operative sensors such as EM trackers, Ultrasound images, pre-operative data. Feature detectors and descriptors will be developed that are able to perform optimal discrimination of different areas of interest and for reducing data dimensions and thus improving the computational performance for intra-operative applications.

Since methods developed in this ESR are trained on multi-modal data, they may be adapted and applied both to intraluminal procedure based on video feedback (i.e. colonoscopy and ureteroscopy) but also to cardiovascular catheterization.

Applicants should have a MSc/MEng (or equivalent) in Engineering, Computer Science, Mathematics or related disciplines. Applicants must have strong programming skills and background in Computer Vision and Machine Learning.

Main  institution and supervisor: UNIVR, D. Dall’ Alba 

Secondary institution and supervisor: UNISTRA, N. Padoy

ESR 12

Distributed follow-the leader control for minimizing tissue forces during soft-robotic endoscopic locomotion through fragile tubular environment

Using prior experience on developing snake-like instruments for skull base surgery, ESR12 will elaborate these mechanical concepts into advanced soft-robotic endoscopes able to propel themselves forward though fragile tubular anatomic environments to hard-to-reach locations in the body. The ESR has two main goals:

  • to use advanced 3D-printing to create novel snake-like endoscopic frame-structures that can be easily printed in one printing step without need for assembly, and that easily integrate cameras, actuators, biopsy channels and glass fibres. FEM-simulations to optimize a) shapes that are easy to bend yet hard to twist and compress (e.g. by using helical shapes), b) minimal distributions of actuators enabling complex and precisely controlled motion.
  • to develop follow-the-leader locomotion schemes for moving through fragile tubular environments (e.g. colon or ureter) and evaluate this ex-vivo in anatomic tissue phantoms.

Main  institution and supervisor: TU Delft, J. Dankelman

Secondary institution and supervisor: POLIMI, G. Ferrigno

ESR 13

Path planning and real-time re-planning

Given the deformable nature of the surroundings, RT planning and control is needed in order to guarantee that a flexible robot reaches a target site with a certain desired pose. ESR 13 will implement an accurate kinematic and dynamic model of the flexible robot incorporating knowledge on the robot limitations right in the planning algorithm so that the best paths are executed

  1. pre-operatively, considering the constraints on allowable paths, the location of the anatomic target and
  2. intra-operatively including the uncertainties in the adopted (and identified) flexible robot model and of the collected sensor readings (ESR5).

Advanced exploration approaches will be adapted to each specific clinical scenario, its constraints as well as the robot constraints such as its manipulability. Specific clinically relevant optimality criteria will be identified and integrated. This methods could try to keep away sharp parts of the instrument (e.g. tip) from lumen edges. RT capabilities will grant the possibility to re-plan the path during the actual operation.

Main  institution and supervisor: POLIMI, E. De Momi

Secondary institution and supervisor: TU Delft, J. Dankelman

ESR 14

Automatic endoscope repositioning with respect to the surgical task

Surgery often involves performing delicate operations with two hands or instruments. Those operations are typically even more difficult when considering MIS done in an intraluminal setting. This is due to the limited controllability of the flexible instruments and to the restricted access to the surgical site. Moreover, humans experience problems in executing complex tasks with flexible multi-arm endoscopes because of the large number of DOFs and the coupling between DOFs. Starting a surgical gesture in an unfavourable position may necessitate an interruption for repositioning, which is not always acceptable clinically.

ESR14 will develop a set of algorithms for intelligent repositioning of the endoscope and its arms in a favourable position. First, the intended gesture is detected using a task model combined with intraluminal sensing. Then, planning and control algorithms are developed to position the system such that the gesture can be optimally performed. While ESR14 aims to develop a semi-automatic robotic repositioning in where robot and operator collaborate. This is a large step towards full autonomy. The main target will be Colonoscopy and Gastroscopy, where robots are usually more complex, but the methods developed will be generic and therefore applicable to other scenarios.

Main  institution and supervisor: UNISTRA, M. de Mathelin

Secondary institution and supervisor: KU Leuven, J. de Schutter


Position has been re-opened

New dead-line: July 21, 2019


ESR 15

Optimal learning method for autonomous control and navigation

Learning optimal control strategies for autonomous navigation and on-line decision making is a challenging problem. Currently 2 main strategies could be adopted: learning from data acquired during the execution of surgical procedure by expert surgeon or learning by experimentation.

ESR15 will compare motion control strategies for intraluminal navigation learned from expert data with the ones learned in simulated environment. From these results, it will be possible to find the optimal control strategies for different clinical scenarios and considering specific robotic configuration. The trajectory planning methods developed in ESR9 will be used to define an initial trajectory for the autonomous navigation. The performance of the different approaches will be evaluated in a realistic setting (physical phantoms) and in a simulated environment using a set of objective evaluation metrics.

An integrated testing environment, including advanced visualization, will be developed to improve the evaluation and testing of different methods (extending/integrating the results of ESR10). The proposed navigation strategies will be tested in the colonoscopy and ureteroscopy clinical scenarios, but possible extensions to cardiovascular catheterization will be considered.

Applicants should have a MSc/MEng (or equivalent) in Engineering, Computer Science, Mathematics or related disciplines. Applicants must have strong programming skills and background in Computer Vision and Machine Learning.

Main  institution and supervisor: UNIVR, P. Fiorini

Secondary institution and supervisor: UPC, A. Casals