Call3 Monitoring AssRob

Public Summary Month 10/2013

Fri, 18 Oct 2013 14:53:20 GMT

9^th Public Report:

The complete robot system is currently being integrated. The mobile column was constructed, manufactured and assembled (see Figure 3). It is capable of carrying the two robot arms, their controllers and powers supplies. The construction is very solid designed (over 250 Kg) and with the centre of gravity being located very low over the ground to prevent it from tilting over. The power supplies etc., the controllers, cameras and emergency chain components have been integrated in the column. Also a switch was integrated to reduce the number of Ethernet cables going from the control pc to the robot column.

Figure 1: Mobile robot column with the location of the integrated components shown.

The new gripper (for the construction see report no. 8) was manufactured (see Figure 1). To allow the retractor to be coupled to the robot, a retractor adaptor was manufactured that also carries the ALVAR marker. The retractor adapter can be mounted on a standard medical retractor. The gripper allows a fast coupling of the retractor.

Figure 2: On the left side the retractor with retractor adapter and ALVAR marker moved towards the gripper can be seen. On the right side the closed gripper with the retractor is shown.

Public Summary Month 5/2013

Mon, 03 Jun 2013 16:53:21 GMT

8^th Public Report:

The complete robot system is currently being integrated. A movable column, capable of carrying both robot arms, their controllers and powers supplies was constructed, most parts have been fabricated in the workshop and the assembly has begun. The power supplies, the controllers, cameras and emergency chain components will be integrated in the column.

New USB-to-CAN interfaces were acquired and the drivers were integrated in ROS framework, excluding the necessity for a second workstation just for interfacing with the F/T Sensor.

A preliminary algorithm was developed to hold the tool after the surgeon couples it to the gripper. It only rotates the retractor around a bone axis keeping the tip fixed, but it showed us that the robot responds well to changes of the external forces, moving the tool in a smooth manner (Figure 1).

Figure 1: Current configuration and movements performed by the robot while holding the retractor (current algorithm only allows rotations around a bone axis).

Public Summary Month 3/2013

Wed, 27 Mar 2013 14:08:18 GMT

7^th Public Report:

An interface between the FT-sensor and Control PC was established, which enabled us to read the output data of the sensor. With that, the Force-Free Guidance mode was implemented. The user is now able to move the robot freely with his hands in a smooth and precise manner. This method enables the surgeon to manually align the gripper with the tool with high precision, as an alternative to the autonomous method (based on the tool tracking algorithm).

For the camera system, we had to calculate the extrinsic calibration. This was done with a flat checkerboard pattern on the base of the robot. The calibration was quite precise; however we will try to assemble a more appropriate marker, but for now it proved to be sufficient.

The approach of the robot to the tool had been previously done in simulation. Now, with the robot ready and camera calibrated, we could finally do it in the real world, with a maximum error of ~1cm. Thus an enhancement of the precision seems to be necessary.

Public Summary Month 1/2013

Thu, 31 Jan 2013 08:44:08 GMT

6^th Public Report

Robot and controller were successfully set up and work as desired now (Figure 1). To improve the robot movement while holding the retractor tests on a phantom simulator on site were performed to evaluate necessary parameters for the control to achieve a smooth movement [Results Submitted, CARS 2013, Gundling et al. Title: Phantom simulator for retractor holding tasks in robot assisted hip surgery].

Fig. 1: Schunk lightweight robot arm mounted on a massive column.

Public Summary Month 11/2012

Wed, 28 Nov 2012 15:04:02 GMT

5^th Public Report

Improvements have been made to the F/T free mode in order for the robot to move more smoothly. Holding mode also needs to be improved since the movement is not smoth.

A collision detection algorithm has been implemented in order to prevent collisions between the robot and the objects/people in the environment and with itself. During simulated tests, the system has proved capable of stopping the robot automatically before hitting objects (Figure 1). Improvements can still be made since the collision map is not updated in real time while the robot is moving.

Fig. 1- Simulation of the Schunk lightweight robot arm approaching an obstacle detected by the camera system.

Although the real robot is not assembled yet finally, most of the control modules are present and have been partially developed and tested successfully in simulation mode, so their integration in a main control program has started.

Public Summary Month 9/2012

Fri, 28 Sep 2012 08:53:58 GMT

4^th Public Report

In tests with a robot system utilizing a phantom simulator that imitates roughly the situation in the OP scene when holding retractors we could identify that it’s beneficial to mount the robot in a higher position. This way it can reach the workspace approaching from above and the robot will less likely get to edge of configuration space (above/below). Further mounting the robot side wards to the column instead should prevent the robot to get into singularities that occurred often in the tests due to alignment of two rotational axes.

The gripper was redesigned to be functional again if robot is mounted differently. In summary these changes reduce the risks of the robot to collide with patient staff. The new setup can be seen in Figure 1.

Figure 1: CAD design of the column for the newly arranged retractor holding robot utilizing the new gripper design.

We started testing markers with chess patterns to track the 6D pose of the tools. These markers were bundled in a cube so they can be seen from every direction and mitigate the problem of partial occlusions. However, a camera setup different than what was previously planned might be needed, due to the lack of precision when the markers are far away from the camera.

The positioning of the eye-to-hand camera relative to the rest of the system has been specified.

To handle the risks brought along with utilizing a robot system medical and technical experts met and discussed what risks have to be handled and how this can be done. A list of system components was created, together with associated safety standards.

Public Summary Month 7/2012

Thu, 26 Jul 2012 13:10:45 GMT

3^rd Public Report

We recieved the robot controller and started working on the control algorithm successfully.

The communication interface between Control PC and Robot Controller was implemented. The control was programmed to command the robot between four different working modes. Since we don’t have all the hardware yet, all tests were conducted in simulation mode (Figure 1).

Figure 1 – The simulated robot holding and rotating the (imaginary) tool around the bone.

We acquired the camera and tested it successfully.

The gripper for holding surgical bone levers was redesigned to be shorter in length and allow to couple medical retractors from the front side.

To handle the risks brought along with utilizing a robot system medical and technical experts met and discussed what risks have to be handled and how this can be done.

Public Summary Month 5/2012

Fri, 25 May 2012 15:22:41 GMT

2^nd Public Report

In tests with the existent assistance robot system and a mechanical simulator developed in the previous “AssRob Project” the alignment of the robot was examined. It was found that the workspace gets increasingly reduced as function of the attached end-effectors length. This result will be considered when optimizing the coupling devices. A further result of the tests was that the flow of forces of the current gripper design results in discharging the less powerful joints of the robot, as planed.

For the tracking of medical retractor by the robot system we decided to use a RGB-D sensor for the pose estimation of the tools, mounted on a fixed location. The tracking method will be based on shape recognition and for that we will attach a rigid-body to the tool.

Public Summary Month 3/2012

Fri, 30 Mar 2012 12:11:11 GMT

1^st Public Report

The setup of the modular robot system has been distinguished in cooperation with the robot manufacturer. To validate that the robot can fulfill the planed task to hold surgical bone levers a model of the experiment setup was created.

Further, suitable Force-Torque-Sensors and the power supplies were selected. A robot control was selected and the software setup of the control-PC has been started. It was discussed what cameras and markers can be used. Mounting the robot to a column has been prepared.

Call3 Monitoring AssRob

Public Summary Month 10/2013

9th Public Report:

Public Summary Month 5/2013

8th Public Report:

Public Summary Month 3/2013

7th Public Report: