The LEGO® MindStorm-based Configurable Telerobotics System (LMCTS)

What is LMCTS?

LMCTS stands for "LEGO® Mindstorm-based Configurable Telerobotics System" laboratoy. LMCTS presents a methodology to analyze and model human supervisory control performance in a highly configurable telerobotics system. The system consists of a semi-autonomous mobile robot, a scalable task environment, and a configurable user interface.

The purpose of this research is to determine the viability of LMCTS toward the analysis and modeling of human performance in a supervisory control environment. A telerobot has been defined as a machine or device which extends an operator's sensing and manipulation capability to a remote location (Sheridan, 1992). Given the advances in wireless technology and the pervasiveness of the internet, the potential benefits of telerobotic systems to reduce safety risks and enhance pleasure are now accessible to a larger segment of the population. In telerobotic systems, an operator receives information about the telerobot through some interface. If the operator receives sufficient information about the telerobot and the task environment so as to feel physically present at the remote site, he is said to have a telepresence (Sheridan, 1992). While it may appear obvious that the more presence that a telerobot elicits, the better the user's performance, empirical evidence does not exist to substantiate that the relationship between presence (or telepresence) and user performance is causal (Welch, 1999).

Researchers at LMCTS examined the effects of different tasks and task environments on human performance through testing in a graduate-level course to instruct cognitive modeling techniques. Specifically, we describe five interactive tasks in which students were required to successfully control the telerobotics system. Models of user performance on each of the tasks were developed by class members. Initial modeling and performance assessment results are reported as part of the work in progress to demonstrate the potential of the research program to assess and model single-user as well as collaborative supervisory control systems.

Equipment:

The LMCTS has following LEGO(r) components:
2 ea. LEGO MindStorm Robotic Command Explorer (RCX 1.0) programmable computers;
2 ea. Infrared Transceivers (range ~2.5 m);
7 ea. 9-volt geared motors (max RPM ~350);
4 ea. Lamps;
4 ea. Touch sensors (calibrated as True/False);
2 ea. Temperature sensors (range -20 to 50 degrees Celsius); and
2 ea. Light sensors (reads reflectance from range 0.6 Lux through 760 Lux).

Other LMCTS hardware components include:
1 ea. X10 Xcam2 wireless camera and receiver set (range ~30 m);
1 ea. Digital PC camera; and
1 ea. Analog security camera.

Exercises:

The course required students to complete five in-class exercises in which students must build and control a telerobotic system to accomplish specific objective(s).

Exercise 1
The first exercise was designed to introduce students to teleoperations. It required the operators to navigate the Surface Control Ship (SCS) shown in Figure 1 around one obstacle and finish with a specified scene in view. The control apparatus for the LMCTS is shown in Figure 2.





Task constraints for Exercise 1 are as follows:
Time constraint: None
Initial Position: Constant
Displays: One (security camera)
Visibility: Unrestricted
Actions Allowed: Move (3 dof)
Communications: operator-to-brick

Two students attempted Exercise 1 and both succeeded.

Exercise 2
The second exercise was intended to teach students to conduct rudimentary search in a remote environment. It required the user to navigate the SCS in searching for a pre-specified object in a timely manner.

Task constraints for Exercise 2 include:
Time constraint: 5 minutes
Initial Position: Variable (placed by instructor)
Displays: One (security camera)
Visibility: Restricted (lights-out conditions)
Actions Allowed: Move (2 dof)
Communications: operator-to-brick

Two students attempted Exercise 2 and both failed due to a lack of time.

Exercise 3
In the third exercise, the students were introduced to an undersea task context. The scenario required a student-controlled SCS to retrieve a disabled remotely piloted vehicle (RPV) from the ocean floor. The RPV, which is modeled by a second RCX, provided simulated distress conditions by emitting a beacon every minute. The landscape for Exercises 3 is shown on the right side of Figure 3.



Task constraints for Exercise 3 were as follows:
Time constraint: None
Initial Position: Variable (placed by instructor)
Displays: Two (wireless and security cameras)
Visibility: Restricted (lights-out conditions)
Actions Allowed: Move (3 dof), grip
Communications: operator-to-brick, brick-to-brick

One student group attempted Exercise 3 but failed under lights-out conditions. However, a second attempt under lighted conditions did succeed.

Exercise 4
The fourth exercise extended the scenario introduced in Exercise 3. This exercise required both RCXs to be maneuvered so that the RPV can rendezvous with the SCS. The control for each RCX was assigned to different students within the group.

Task constraints for the fourth exercise include:
Time constraint: None
Initial Position: Variable (placed by instructor)
Displays: Two (wireless and security cameras)
Visibility: Restricted (lights-out conditions)
Actions Allowed: Move (SCS 3dof, RPV 2dof), grip (SCS)
Communications: operator-to-operator, brick-to-brick, operator-to-brick

One student group attempted Exercise 4 twice but failed both times. The failure was due to an inability to sense that the RPV was securely acquired.

Capstone Exercise (Exercise 5)
The Capstone Exercise required a collaborative effort between two teams of students to control achieve two objectives. One team controlled the RPV while the other maneuvered the SCS. The RPV operator was tasked to retrieve an object on the ocean floor. Once retrieved, the RPV operator needed to move the object in position to be retrieved by the SCS operator. The collaborative environment is shown in Figure 4.



The student team maneuvering the SCS used C while the team controlling the RPV used A. The operator at A had access to the wireless and security cameras while the operator at C viewed the environment through the PC camera. Videoconferencing software was loaded on B to enable teleoperations from C. Communication between the two teams occurred through a shared audio link.

Task constraints for the Capstone Exercise include:
Time constraint: None
Initial Position: Variable (placed by instructor)
Displays: Three (wireless, security, and PC cameras)
Visibility: Restricted (lights-out conditions)
Actions Allowed: Move (SCS 3dof, RPV 2dof), grip (RPV)
Communications: team-to-team (remote-local), brick-to-brick, operator-to-brick

Two successful attempts were made for the Capstone Exercise.

Implications of task performance toward existing research:

We found that:
depth and elevation judgments tend to be more accurate in stereoscopic displays (Yeh & Silverstein, 1992). Though stereoscopic vision was not provided multiple cameras increased the user's ability to navigate the environment.
users gained more information in Exercises 3, 4, and the Capstone, which were higher in fidelity than Exercises 1 and 2. Nash et al. (2000) reviewed research that suggest adding navigational aids may provide more desirable means of wayfinding.
while aiding was not added to the interface, operators used landmarks in the simulated ocean environment to navigate.
One main complaint of students was that the mapping between the controls in the NQC-based software package did not correspond to real-world controls. This may have contributed to the failures in Exercises 3 and 4.
Task consistency was often interrupted by technical difficulties in the LMCTS and hence may have hindered operator presence. Moreover, the serial nature in which the exercises were executed precluded motion for multiple components in the task environment. Existing research suggests that this factor may have reduced operator presence as well (Witmer & Singer, 1994).
In terms of communication between the operator and the telerobot in virtual environments, Heeter (1992) found that the equipment should be as unobtrusive as possible. In our experience, we found that as students are hurrying to finish building the RPV or coding supervisory control modules, equipment - often unintentionally - impede successful teleoperations. Nevertheless, the practice undergone by the students as well as their motivation to undertake the task overcame other obstacles (e.g., Nash et al., 2000) and they were able to successfully complete the Capstone Exercise.

Implications toward future research:

From our analysis, LMCTS can be used as a tool to assess user performance and presence in multiple task environments. We also propose that for LMCTS to be fully effective, modifications should be made to the existing system.
The control interface should be redesigned in a more user-centered manner. The existing design hinders effective control and does not map to real-world controls.
The environment should be made more dynamic to reflect the real-world system on which the scaled model is based. This would also likely increase user presence.
The datalogging capabilities of the existing system - which is limited to 6K - must be increased in order to quantitatively assess performance and presence.
The existing video signals should be combined with computer-generated graphics to overcome the artificialities inherent in a scaled model. This type of systems is also known as augmented reality (Pretlove, 1998).

Acknowledgements:
We would like to thank Dave Baum for his advice in the use of LEGO MindStorm sets, and Craig Harvey for sharing his resources to make the class a reality. We also thank the graduate students who attended the "Quantitative Methods in Cognitive Modeling" course. Their enthusiasm and desire to learn were encouraging

References:

Heeter, C. 1992. "Being there: The subjective experience of presence." Presence: Teleoperators and Virtual Environments 1, No. 2: 262-271.

Nash, E. B.; G. W. Edwards; J. A. Thompson and W. Barfield. 2000. "A Review of Presence and Performance in Virtual Environments." International Journal of Human-Computer Interaction 12, No. 1: 1-41.

Pretlove, J. 1998. "Augmenting reality for telerobotics: unifying real and virtual worlds." Industrial Robot 25, No. 6: 401-407.

Sheridan, T. B. 1992. Telerobotics, Automation, and Human Supervisory Control. The MIT Press, Cambridge, MA.

Witmer, B. G. and M. J. Singer. 1994. "Measuring presence in virtual environments." Tech Report 1014. U.S. Army Research Institute, Washington, D.C.

Yeh, Y. Y. and L. D. Silverstein. 1992. "Spatial judgments with monoscopic and stereoscopic presentation of perspective displays." Human Factors 34: 583-600.