How would your lab automation devices sense their environment?
Consider now the sensors that our laboratory machines would need to be equipped with. Given its function in experiments and testing, the robot must be able to sense its environment with considerable acuity. As a minimum the robot must, firstly, be able to sense where and how far away, relative to itself, other devices are. The robot must be able to hear her and localize the source of her voice, turning its head if necessary to face her in response to a verbal request for attention. Second, the robot should be able to see and track its posture, including, importantly, her head orientation and position of her arms and hands in order to ‘read’ her body gestures. Third, the robot should be capable of seeing its mistress’s face with sufficient resolution to be able to recognize her, and her facial expressions; and, since humans use eye gaze to direct attention, the robot must be capable of gaze tracking.
The sensor technology to meet these requirements with some degree of sophistication exists already. Indeed, the body- and voice-tracking technology has already been developed for the interactive video games market in the shape of the Microsoft Kinect device, and it’s no surprise that engineers have been quick to experiment with this low-cost technology. A robot device should also be able sense its general health and well-being, so additional sensors to monitor basic life signs, such as temperature and breathing, would also be a good idea. Even though it is not a nurse robot, a capable robot companion should surely be able to sense if there is something seriously wrong with its mistress and call for help. Of course, although our robot companion’s senses need to be focused on its human mistress, the robot also needs to be able to sense its environment and the objects in it with sufficient fidelity to be able to safely move around and through it. To support its valet functions, the robot must be able to sense and recognize objects well enough to be able to handle them, including handing them to its mistress (food or drink), or co-handling objects with its mistress (clothes, for instance, while assisting with dressing or undressing).
How would your lab automation device interact?
Let us now turn to what is undoubtedly the major technical challenge we face in designing our robot companion: its artificial intelligence. At the outset I specified the robot as requiring human-level intelligence and, of course, in order to have a conversational capability equivalent to a human butler/ companion (witty, cheerful, well-informed, and diverting) our robot would indeed need human-level intelligence or a convincing emulation. Unfortunately, setting out a plausible shopping list of software ‘modules’ in the same way as I have done for the robot’s sensorium is simply not possible. We just don’t understand the architecture of human intelligence well enough to be able to construct a reasonable artificial emulation. We can split the robot’s AI into two major sections: one for its object recognition and manipulation and the other for its conversational human—robot capability.
The trick to making a good strong joint is to create a joint that fits well and has a thin continuous layer of glue in it. It is not a good idea to close the pores of the wood with firm wood dust because open pores help the glue to get into the rod and this makes a better glued joint. An attended joint will make a better glued joint if you make sure that the joint is well fitted before you apply the glue. Make the outside parts over thin and then sand the parts to size after they are glued together. This will give you the best finished parts. Use a coarse paper to get parts to stick when fitting them together. Coarse sanding will not fill the pores with fine wood dust the way sandpaper will. Hand sand the finished joints prior to painting to kill the edges and get a neater appearance.
Setting the Motion
Most of the photographs on this website show painted components to allow you to see the components in greater detail. Once the two casings for the foot servos have been built, build the soles of the feet. These should be a firm, but shake free, fit on the feet. One servo should be able to move the foot without binding. On this one joint, the range of motion needed is quite limited because even a sway motion of the foot will tip the robot considerably. Make the servo lever as short as possible and the foot lever as long as possible. The motion should be shake free so that the robot does not stutter on its foot. What is needed is a small jaunt of very backlash free motion so that the control can be positive as well as responsive. You might want to consider making the soles of the feet oversized in the beginning and then strip them back later. This will make for a more capable robot while you develop the skill to control them. While each section of the legs is built, make sure t each component can move through its entire range without binding. Each joint is made identical of all other joints so no complicated fit procedures are needed.
Setting the Movement
You should try to adjust the movement you need around the middle to allow for 90 degrees of motion of the servo. Adjust rod lengths and crank lengths accordingly. After the feet have been completed, build the calf sections and then the thigh sections of the legs. These are similar to one another and are straightforward. Since these parts are wider than the foot section, there is now more space to be filled with spacer. Motion is limited by using the actuation arms on the servos at their shortest adjustment. This can be increased later. The right leg of the robot as seen from the right. Note that the lengths of the servo arms are adjustable. Keep them short to start with and then adjust them as needed by the software that you develop. All linkages are on the outside for ease of adjustment and removal. Note the bearing reinforcements at each joint. The left leg showing a bend at the knee joint while the hip stays above the ankle.
When it comes to robots, as human beings we love a robot fashioned in our own image. Science fiction has pushed this idae of what a robot should be, almost to the exclusion of all others. So it’s no wonder that anytime a robot builder creates a robot with two legs, two arms and a head, the world beats a path to the laboratory door. The world has been so taken by Sony’s humanoid robots that very few people (including otherwise skeptical journalists) have bothered to ask what sort of brain control these robots are capable of. This is throughout all robotic applications and lab automation is no different. Sony spent over one hundred years of man hours desperately trying to develop a robot in this image. They spent a significant amount of this time just getting the movement of the robot correct before even looking at the robots functionality. It was also diffuclt trying to figure out how to get all the electronics and servo motors into a human shaped body. The products created were really prrof-of-concept models for automation mobility.
Do not get me wrong, there is nothing wrong with methodolgy at all. As can be seen in many lab set-ups, a bottom-up approach makes a lot of sense when it comes to building an automated machine. With most of these humanoid robots, a human oeprator is actually instructing the robot, to perceive a sense of artifical intelligence. This is a way of masking the defficiencies of the robot and make the over all project look more succesful than it actually is. Robot makers from all over the world and the world’s humanoid research labs, are thrilled that these bipedal bots have so thoroughly captured public’s imagination.
The latest models from companies like Honda and Sony, are all part of a country wide humanoid-robot initiative in Japan. Here robot researches are working hard trying to creat autonomous bipedal robots that can interact sucessfully with every changing environment and perform the most complex of operations. The idea behind this project is also to create automated machines that are not essentially applicable to all life scenarios but can atleast handle multiple tasks. The realization of this particular vision may not be as far-off as once was thought.
This is a very exciting time for me and hopefully for you to as we strive to break down the barriers of all things robotics. Automation already plays a huge role in our every day lives but this is set to increase significantly in the coming years.
robotbin 