Team 31 CART Update (11/15/2019)
Thanks for reading our blog! If you haven’t seen it already, you may want to check out our first blog post.
Introduction and Problem
As discussed in our previous post, our challenge lies in designing an effective ‘short-distance’ transportation system for visitors and students/faculty unable to walk around the University of Houston campus. A potential solution lies in using golf carts to navigate between points of interest on campus. Golf carts are ideal because they are small enough to drive on sidewalks and paths that are inaccessible to regular vehicles. Further, by making these golf carts autonomous, there is no need for a dedicated driver, so the University can reduce the operating costs of this transportation system. Eliminating the human driver reduces cost while also increasing the safety of cart, as more than 90% of crashes are caused by human error (NSC 2019).
Unfortunately, there are no commercially available autonomous golf carts. Despite multiple attempts (e.g. Auro Robotics, which pivoted to autonomous cars), no company has been able to deliver a fully-autonomous golf cart, let alone at a reasonable price. The Cougar Autonomous Robotic Transport (CART) team proposes to modify an existing golf cart (which is called the CART) to be capable of level 3 autonomy as defined by the Society of Automotive Engineers (SAE). This means that the CART will drive itself, but requires that a human driver be present to take over in rare situations where the CART doesn’t know how to proceed. We plan to build on the work of a previous capstone team that purchased a used golf cart and installed actuators (steering, brake, and accelerator) to control this golf cart with a remote controller.
Progress during Nov 1st - Nov 15th
Since our last post, we conducted trade studies to determine which sensor units we will use. We decided on a LIDAR unit from RobotShop.com because it provides the best balance between performance and ease of integration into our system for our budget. We plan to use the LIDAR for mapping and obstacle detection. RADAR and SONAR sensors are both cheaper and theoretically feasible options, but they aren’t great at mapping, and perhaps most importantly, are difficult to implement for our use case with our skillset.
We also decided on an XL-MaxSonar-EZ ultrasonic unit from MaxBotix.com. These inexpensive units provide adequate range and resolution to detect obstacles very close to the CART.
We are still looking at our options for implementing an odometry sensor onto the CART, although we have not decided on a specific unit yet. It is difficult to access the drive axle on the CART, so we are still determining the best location for a sensor to obtain wheel speed information.
Another design item we completed was our overall system architecture. Note that we still need to design specific items such as our wiring diagrams but our high level information flow has been designed. The following picture is a good summary of this architecture.
As you can see, the various sensors feed information about the cart and its environment to the main mapping computer and the autopilot computer. This information is then processed into driving commands, which the autopilot computer sends to the actuators.
Lastly, we cemented our plans for the upgrades to the steering and braking systems. We decided to continue using the same steering gear that came on the cart, but after we press it back into position, we will have it welded in place to prevent it from coming loose again. For the brake upgrade, we decided to increase the pressure in the compressor and run a second, high pressure line to the brake piston. We decided to go this route because the cost was on par with our other options, and this option required the least invasive modifications to the existing system.
Notable challenges encountered
As mentioned above, we plan on adding some sort of odometry sensor to the CART. Odometry is used to determine how fast something is moving - your car has one that displays your speed on your dashboard. The CART currently has nothing like this. We’d like to add one because the data can be used as a supplement to GPS data to better determine the CART’s location.
The simplest odometry sensor adds a magnet to the wheel, and has a sensor that detects when the magnet passes it. As the wheel spins, the magnet passes the sensor multiple times, and the sensor records the number of passes. By doing this, rotations per second can be counted, and speed of the vehicle can be calculated from this.
This may not be possible for us. The CART has very limited space where we’d like to mount such a system, so we’re unsure if we have enough room. Therefore, we may have to obtain odometry using some other sensor. We are currently looking for the best solution that is cheap yet reliable. In the worst case that we cannot determine odometry mechanically, there is a method to obtain odometry via electrical current draw, although this has limited accuracy.
Our budget challenge is more subtle. While conducting trade studies, we noticed that the more expensive sensors were typically safer, more reliable, and better supported in software than the cheap sensors. Safety is our biggest concern, so we don’t wish to compromise the safety of our project by selecting cheap (and unreliable sensors). By increasing our budget, we hope to increase the safety and reliability of the CART.
We are actively addressing our budget and workspace constraints. To address our workspace challenge, we are waiting to hear back about a potential CART workspace provided by the mechanical engineering department. This workspace would be conveniently located on the University of Houston campus, thereby allowing easy access to the CART for testing and modification. We are addressing our budget challenge by compiling a list of professors and other potential sponsors for our pricier budget items. We hope to offset some of our perception sensor costs by collaborating with IEEE Makerspace and by purchasing sensors at a discounted rate.
Work plan for Nov 15th to Nov 25th
In the near future, we hope to finalize our system architecture. This means that we’ll make those more detailed wiring diagrams and draw up the details of how our software will work. One potential issue with presenting these plans is that we won't know exactly how our software works until we finish writing and testing it. We currently have a design outlined, but we expect that we'll have to fine-tune some things along the way to optimize the CART, so our final implementation probably won't look exactly the same as planned.
We also hope to reach out to professors on campus to see if any would be interested in advising us. We anticipate that weird issues could occur with using our sensors that an experienced professor might know how to solve. Note that we do not plan to install sensors before November 25th, but we'd like to know ahead of time which professors we need to go to for advice, should we encounter issues with our sensors.
We plan to start ordering things and fundraising, although we may not get to fundraise in earnest until winter break. Unfortunately, the CART team is embroiled in the very busy end-of-semester academic season, so we have less time than we’d like to start a strong fundraising campaign.
We plan to start ordering parts very soon. Besides the LIDAR, we currently have the funds to purchase the rest of our items from our budget outlined above. It is a good idea to order components now, because if we order them now, we should receive them in time for winter break.
Introduction and Problem
As discussed in our previous post, our challenge lies in designing an effective ‘short-distance’ transportation system for visitors and students/faculty unable to walk around the University of Houston campus. A potential solution lies in using golf carts to navigate between points of interest on campus. Golf carts are ideal because they are small enough to drive on sidewalks and paths that are inaccessible to regular vehicles. Further, by making these golf carts autonomous, there is no need for a dedicated driver, so the University can reduce the operating costs of this transportation system. Eliminating the human driver reduces cost while also increasing the safety of cart, as more than 90% of crashes are caused by human error (NSC 2019).
Unfortunately, there are no commercially available autonomous golf carts. Despite multiple attempts (e.g. Auro Robotics, which pivoted to autonomous cars), no company has been able to deliver a fully-autonomous golf cart, let alone at a reasonable price. The Cougar Autonomous Robotic Transport (CART) team proposes to modify an existing golf cart (which is called the CART) to be capable of level 3 autonomy as defined by the Society of Automotive Engineers (SAE). This means that the CART will drive itself, but requires that a human driver be present to take over in rare situations where the CART doesn’t know how to proceed. We plan to build on the work of a previous capstone team that purchased a used golf cart and installed actuators (steering, brake, and accelerator) to control this golf cart with a remote controller.
Progress during Nov 1st - Nov 15th
Since our last post, we conducted trade studies to determine which sensor units we will use. We decided on a LIDAR unit from RobotShop.com because it provides the best balance between performance and ease of integration into our system for our budget. We plan to use the LIDAR for mapping and obstacle detection. RADAR and SONAR sensors are both cheaper and theoretically feasible options, but they aren’t great at mapping, and perhaps most importantly, are difficult to implement for our use case with our skillset.
We also decided on an XL-MaxSonar-EZ ultrasonic unit from MaxBotix.com. These inexpensive units provide adequate range and resolution to detect obstacles very close to the CART.
We are still looking at our options for implementing an odometry sensor onto the CART, although we have not decided on a specific unit yet. It is difficult to access the drive axle on the CART, so we are still determining the best location for a sensor to obtain wheel speed information.
Another design item we completed was our overall system architecture. Note that we still need to design specific items such as our wiring diagrams but our high level information flow has been designed. The following picture is a good summary of this architecture.
 |
| System architecture for the CART |
As you can see, the various sensors feed information about the cart and its environment to the main mapping computer and the autopilot computer. This information is then processed into driving commands, which the autopilot computer sends to the actuators.
Lastly, we cemented our plans for the upgrades to the steering and braking systems. We decided to continue using the same steering gear that came on the cart, but after we press it back into position, we will have it welded in place to prevent it from coming loose again. For the brake upgrade, we decided to increase the pressure in the compressor and run a second, high pressure line to the brake piston. We decided to go this route because the cost was on par with our other options, and this option required the least invasive modifications to the existing system.
Notable challenges encountered
As mentioned above, we plan on adding some sort of odometry sensor to the CART. Odometry is used to determine how fast something is moving - your car has one that displays your speed on your dashboard. The CART currently has nothing like this. We’d like to add one because the data can be used as a supplement to GPS data to better determine the CART’s location.
The simplest odometry sensor adds a magnet to the wheel, and has a sensor that detects when the magnet passes it. As the wheel spins, the magnet passes the sensor multiple times, and the sensor records the number of passes. By doing this, rotations per second can be counted, and speed of the vehicle can be calculated from this.
This may not be possible for us. The CART has very limited space where we’d like to mount such a system, so we’re unsure if we have enough room. Therefore, we may have to obtain odometry using some other sensor. We are currently looking for the best solution that is cheap yet reliable. In the worst case that we cannot determine odometry mechanically, there is a method to obtain odometry via electrical current draw, although this has limited accuracy.
 |
| Minimum feasible budget |
As mentioned in our previous blog post, we are severely constrained by our budget and our available workspace. Without a workspace, we are unable to make any significant modifications to the cart. As we proceed with our project, we need a workspace to test subsystems and to reduce the risk of inadvertent damage through theft or accident.
Our budget challenge is more subtle. While conducting trade studies, we noticed that the more expensive sensors were typically safer, more reliable, and better supported in software than the cheap sensors. Safety is our biggest concern, so we don’t wish to compromise the safety of our project by selecting cheap (and unreliable sensors). By increasing our budget, we hope to increase the safety and reliability of the CART.
We are actively addressing our budget and workspace constraints. To address our workspace challenge, we are waiting to hear back about a potential CART workspace provided by the mechanical engineering department. This workspace would be conveniently located on the University of Houston campus, thereby allowing easy access to the CART for testing and modification. We are addressing our budget challenge by compiling a list of professors and other potential sponsors for our pricier budget items. We hope to offset some of our perception sensor costs by collaborating with IEEE Makerspace and by purchasing sensors at a discounted rate.
Work plan for Nov 15th to Nov 25th
In the near future, we hope to finalize our system architecture. This means that we’ll make those more detailed wiring diagrams and draw up the details of how our software will work. One potential issue with presenting these plans is that we won't know exactly how our software works until we finish writing and testing it. We currently have a design outlined, but we expect that we'll have to fine-tune some things along the way to optimize the CART, so our final implementation probably won't look exactly the same as planned.
We also hope to reach out to professors on campus to see if any would be interested in advising us. We anticipate that weird issues could occur with using our sensors that an experienced professor might know how to solve. Note that we do not plan to install sensors before November 25th, but we'd like to know ahead of time which professors we need to go to for advice, should we encounter issues with our sensors.
We plan to start ordering things and fundraising, although we may not get to fundraise in earnest until winter break. Unfortunately, the CART team is embroiled in the very busy end-of-semester academic season, so we have less time than we’d like to start a strong fundraising campaign.
We plan to start ordering parts very soon. Besides the LIDAR, we currently have the funds to purchase the rest of our items from our budget outlined above. It is a good idea to order components now, because if we order them now, we should receive them in time for winter break.
