Camera Chase Vehicle

Quad-rotor drone shots taken low to the ground are difficult: GPS altitude is fairly rough on accuracy, and obstacle avoidance can get significantly more difficult versus just flying over the everything. Cinema rover drones are less common but do get around a number of these problems, especially if the subjects are not high off the ground. As a fan of karting, lightweight contraptions and photography this seems like a pretty good project: Tackle a mechanically stabilized video platform, pilot it remotely and capture some outdoor action. Taking a look at the Chassis I picked up this from BMI surplus [link], during an adventure with Jake Hecla [link] and the mysterious Arsenio [link]. For some quick back-story, BMI surplus is this incredibly interesting surplus emporium in Massachusetts, while they do not have tours, you can pick up items that you purchase ahead of time. We were fortunate to get some time to browse inside the incredibly dense arrays of machines, gadgets and gizmos. A lot of this stuff appears to come from Lincoln Laboratory, the friendly neighborhood spooks. I was able to haggle a bit and purchased the mystery RC chassis for ~50 USD + tax, along with some other items. There was no actual information about this thing, what it was used for, why it had a bizarre Z-Axis linear actuator bolted to the top. I did some digging but was unable to find any write-ups, technical papers or anything about this 'thing'. Maybe they were surveying, maybe it was a mobile device to take images at the door height of a car? No idea Fortunately the actuator was a simple DC brushed motor linear actuator, which was easy to control. Behold a scaled dolly shot of this contraption in-action. There is an animated video here tag. While this thing was interesting, there was no way that I was going to even try putting a full sized camera gimbal on this thing, so after a few screws the Z-Axis was removed, underneath we see a water-jet plate on, the most comically dubious 4-40 standoffs I have ever seen. 4 inch long 4-40 stand-offs is wild.

Hardware Mock Up For the purposes of determining what this could look like, and if it was going to be too unwieldy, I opted to temporarily use the existing mounting plate, make a 3D printed adapter and get a hint as to what I'd be working with. First up was to remove the handle assembly that's native to the Movi M10, which inherently flips this whole gimbal upside-down. I was somewhat concerned with how much loading would now be on these four M3 screws, but that's something that future-Dane needs to resolve. Shown below is the gimbal, with the mount point shown. Likely, going forward I would probably end up picking up the central 4 screws and try and mechanically provide some additional support. I used some quick thermal inserts to pick up on the long flat portions of the existing chassis. By using flat-head screws, I can take up a bit of tolerance mismatch, and when loose it can slide back and forward down the top platform. This print was a placeholder, but it did let me get a good visual of how tall this whole stack-up would be. Finally, the first test fit to see what the gimbal looked like on the chassis. While this is a quick mock up it did provide some insight as to how quickly the height of the whole assembly could easily creep up if it wasn't actively being constrained. The lower the camera and assembly height, the lower the effective vehicle center of gravity remains, increasing its stability while helping reduce rollover forces. Making a mock-up may take time but it really lets you skip an iteration step as instead of an ill-defined cad model you now have something to interact with on the lab bench. There is an animated video here tag. It is huge. I took the opportunity to also do some placement tests for where the battery mounts could end up. As it was fairly apparent, using batteries as bumpers is not a great plan. The two packs would likely need to lay as low as possible but still remain accessible for hot-swapping two packs at a time. The only reasonable spot would be along the sides, while not interfering with the gimbal or the ground clearance. New Mechanicals A more structural gimbal mount The floppy aluminum mounted on some adorable 4-40 standoffs was not going to do it, I needed something much more structural.

While the gimbal itself is not mechanically heavy, its mount point does need to be as low as possible to help mitigate flipping. Ideally the Gimbal-camera assembly is remove-able from the frame, such that i can test and tweak the motor control tuning, fenders and the like without putting everything else through hell. Time to fire up the CNC, turn some proper standoffs and rigidly connect to the frame. The basic plot is to provide a solid foundation as close as mechanically possible to the frame for the gimbal and shock mount. Fortunately the sub-frame base plate is made of aluminum. I'm going to again opt for stand-off's to elevate the platform above the drive motor, but just use some large round-stock to provide a secure mount. There is an animated video here tag. After some quick machining, drilling and tapping we have our new elevated gimbal platform, with countersunk M8 flat head screws. This intentionally barely clears the drive motor, and leaves the central round part of the spring damper mount recessed to keep the Z-offset height as low as possible. The spacing of the standoffs is nominally tied to the closest positions that I could pick up that were co-planar without bumping into the motor mount / servo mounts. While this is slightly aft of center, it does permit space for the somewhat heavy batteries to live, ideally resulting in the total mass balance towards the center. After verifying location a few times, I punched some holes through the chassis and used some M8 screws with flanges to firmly attach the new platform to the chassis. These were initially just tightened to a few newton meters, but on the final assembly did receive some mild loctite to help with vibration induced loss of tension. Fenders for Ice Racing One of the big issues with racing on a slushy surface is the slush getting everywhere . I do not have a traditional chassis for this vehicle, so it's up to me to figure out how to contain the slush, while not being too inflexible. This is a great option for 3D printed parts, as it's a lot of odd shapes and contours, however, this is also a battle bots crossover episodes , and oddly, flexible things should out-perform static things. Initially, for iterating and quick prototyping, the fenders are normal PLA.

Comically, up to this point, I have never actually printed with commercial TPU, which is the go-to plastic flexible elastomer. I cant think of a better use case than fenders on a RC car frame. Just like a commuter-bike, the more of the wheel path that is covered, the less can get sprayed onto the camera and gimbal. The plot is to pickup a hard mount, which in this case is some aluminum angle stock, use a standard PLA part to provide a mounting point close to the wheel, and then transition into a TPU part around the wheel. I had initially tried 85A TPU and it was just a bit too fiddly to reliably print, without re-doing my filament spool holders to be lower resistance. Any resistance was causing under-extrusion and it was difficult to manage. I opted for switching to 95A filament, which is stiffer, and did a subsequent slow weekend print to make flexible tire fenders for the rear. The rear mounts are mirrored but both mate with three M6 threaded flange screws. After fiddling with settings to get a reliable print on the fenders, its a good idea to start planning a TPU front bumper to help keep this thing from getting smashed too easily. The print time for a single wheel fender was approaching a day and a half on a Prusa MK4. For this specific print, I opted for high wall count 25% infill, this should result in a fairly stiff part that's able to absorb impacts. While it would have been ideal if all four fenders were the same part, the steering up front requires a lot more clearance, resulting in a larger radius. The printed part does pick up the same hard mount point There is an animated video here tag. I didn't mention how awkward removing the support material is for large TPU prints. Its time consuming just due to how impact and force absorbent it is. The layer-layer adhesion is amazing. There is an animated video here tag. Finally a spool up of the chassis on stand-off blocks, fueled by the two series DeWalt batteries. As I learned from a friend, full RPM in this situation is outside of the motor and drive-train specs, given that i was now running 10S / ~40v. The short blip of full speed was plenty to see how frightening this monster would become.

There is an animated video here tag. To mount the side plates, i used a long tip marker to indicate where the print would align with the chassis and used a punch to transfer those locations so they could be subsequently tapped. There is an animated video here tag. With the holes in the gimbal riser tapped and the chassis clearance holes drilled, it's time to put everything together. Due to the right side of the vehicle's simplicity, I'm opting to install its cover plate first, there's only two wires to worry about. Time to put the electric screwdriver to work. Six M3 screws grab the aluminum top-plate and six subsequent M3 screws attach the bottom of the print to the frame. There is an animated video here tag. This video has audio: Click to un-mute Battery Mounts Both sides of the gimbal mounting plate fit in separate, blue, printed plates that hold the main power switch, pre-charge and battery mount points. The prints pick up tapped M3 holes above and contain M3 thermal inserts below to pickup screws from the chassis. The printed mounts then mate to an off the shelf, injection molded DeWalt Battery terminal, connecting with four M4 screws. With the battery latched in, it is surprisingly abuse tolerant. Shown below is a standard DeWalt 6AH 20V battery module. The orange cover captures any exposed wiring present from the battery adapter. Ideally the gap-space gets covered up to help mitigate ice slush ingress, but that's a problem for future Dane. There is an animated video here tag. For the left side of the chassis cover we have a lot more going on. The main power switch, pre-charge and pack voltage indication is present, with associated wiring on the backside. Behind this are the dc/dc converters that provide steering power and indication light power, along with our remote controlled relay for headlights / tail lights. "Headlights and Tail lights" Having some visual indication on this contraption is helpful, especially with how quickly dusk appears. A small headlight and rear facing red lights should be a quick addition: For a headlight i opted for this waterproof small module, intended as a 3rd party automotive light. It fortunately takes a wide range of operating voltages so i can put to use a 15W dc/dc module that's been collecting dust.