Project conceived and built with Patsy J. Gomez
Summary
The Zemiscua robot was created to fulfill the prototype-thesis requirement to obtain the B.Sc. in Automation Engineering at Universidad de LaSalle in Colombia.
It was designed to serve as a mobility base for other robots in uneven terrains with obstacles smaller than 10cm.
The story behind the robot and the friendship
My friend and colleague Patsy and I were in the second group of students ever admitted to the Automation Engineering program. It was the first of its kind in the country (Colombia) with the characteristics of a Mechatronics (sort-of) program. At the time, LaSalle was the only university willing to take the risk on such a program. And we were willing to be guinea pigs. This meant that there were no automation engineers to lead the way. We were making history (OK, not that much grandeur, but still, it sounds cool).
Because it was such a new program we were also the only school that saw close to 40% of women applying and admitted. It was the '90s, so it had not been seen before in an engineering program related to electronics or mechanics. And as far as I know it never happened again (please support women in STEM).
Patsy and I were equally nerdy, and after finding each other we became besties and together we presented almost every group homework and took every class together, as we could. As much as we could. The common interest on Artificial Intelligence, Robotics and the Environment were the glue to our friendship.
When the time came to present a proposal for a prototype, there were just not proposals from the first admitted. "Natural Selection" and tons of Calculus were to blame on the progress "delay". Because of this the students of my class were the first.
No one knew how the prototype should look or what the scope should be. Naïveté took care of setting the goals and we were to fulfill the scope, without the required "interdisciplinary" guidance.
Because the brain of our project was an Expert System, the prototype committee chose as project director the part-time Artificial Intelligence professor. Artificial Intelligence was the least of our problems. We had mechanical and electronics problems to be solved and very few people to ask, but mostly no robotics expert.
It took a long time, but we eventually made decisions, sometimes not the best ones. We made decisions without trying to over burden our parents. We were fortunate that family and friends pitched in to help. We were grateful for their help. It enabled us to develop the skills we now have.
At the time NASA's mission was venturing on Martian soil and the similarities to the mission's robot earned the Zemiscua the nickname of Path Finder among teachers and students.
The pressure on the dean to graduate students translated on pressure on students to finish projects. With a deadline, Patsy moved to my apartment, where we had an improvised workshop and worked 6 days a week, 14 hours a day for more than 3 months.
We made it! Even with issues like being one week away from our thesis presentation and the hard drive catastrophically failed. I would not change that experience for anything. Looking back, it still remains one of the happiest times in my life.
Why "Zemiscua"? you may ask. We wanted a catchy name that was local and meant "messenger". An anthropologist friend of my dad, gave us the name in Muisca language, local to the indigenous of the Bogotá region. We assumed it was right, at the time was really hard to fact check it.
Please forgive minor errors while reading the technical stuff. It has been a long time, technology has changed so much, and the engineers working on the project have learned a ton.
The Technical Stuff
The robot Zemiscua is an autonomous navigation platform that provides mobility to a "user robot" with the ability to avoid obstacles on the road. The obstacles are classified as surmountable, if their height is less than 10 cm, and non-surmountable, otherwise.
5 subsystems make the robot work:
Motion and structure
Power
Information acquisition
Decision making
User Interface
Motion and Structure
Three main elements are part of the motion system:
Chassis: is a rectangular platform with dimensions 70x80cm. It supports the physical structure of the robot, the traction system and the sensors, power, interfaces and the cover
Traction: 4 DC motors and the electronics and mechanics plus 4 tires of 12in in diameter. The max speed is 5.5m/min. The turns of the platform to the right and left are achieved by blocking the tires on the opposite side.
Cover: It is made of fiberglass by a car body artist. It gives protection to the electronics and gears.
Power
Two car lead-acid batteries (12V-43Ah) and two dry batteries (12v-7Ah) power the robot. The electronics for the motors, sensors, computer main board, peripheries and other circuits also belong to the power system. They provide an operation time of 20 minutes.
Information acquisition
The following parts belong to this system:
An ultrasonic sensor with analog output and 70cm of scope that does a radar-like movement to map the obstacles
5 optical switches that tell the robot when an object is too close
2 bi-polar mercury sensors for inclination detection for the axes x and y of the robot
Motors and controllers for the leveling mechanisms.
These sensors indicate the angle of inclination of the robot, distance to the objects and hollow/gap spaces greater than 10cm, and obstacles behind the robot (when it has to back up).
All this information is transferred to the Fact Base in the Decision-Making System to make decisions.
The sensors have the following limitations:
Scope
Ultrasonic sensors have detection problems with flat surfaces that are not perpendicular to the emission cone. That is why the test objects were covered with a gravel material. If the budget allowed it would have more accurate to have more ultrasonic sensors
Decision-making
The brain is an expert system, a traditional technique of artificial intelligence that uses the knowledge of a human expert translated to code, for the decision-making process.
To make a decision (output), the "analysis system", called Inference Engine, evaluates the current state of the system, called Fact Base, using the "rules" that a human expert would use and that are stored in the knowledge base. The rules have the general shape of "If this and this then do this".
The expert system was built on C++ running on a computer with an 80386 processor.
User Interface
To emulate the instructions given by a robot, a digital keyboard was connected to the processing system. The communications protocol was developed for this project. This also allows the user to visualize errors when identified by the Expert System.
Subsystems block diagram
The following schematic shows the relationships among subsystems.
Applications
Since the project was conceived as an auxiliary robot to provide mobility to a main robot, applications like environment mapping, mine detection, very high or very low temperatures, radioactive, caves and sewage environments would be a great application for the robot.
Current status: dismantled.
Comments