A Ten Kilogram Space Station
The International Space Station cost approximately 150 billion dollars to build and operate as of 2014; assuming an average crew of 5 and 15 years of operation, that is about 2 billion dollars per astronaut-year. Assuming 4000 hours per year of useful activity per astronaut, that is 500,000 dollars per hour. While many tasks require an astronaut on site, many can be done by robots.
Humanoid robots controlled from Earth with predictive-adaptive telepresence can accomplish most tasks that astronauts do, and many that cannot be done by humans. With deep miniaturization, tiny robots can do microscopic tasks very, very cheaply.
Imagine a humanoid robot (a head with eyes, two articulating arms, a base that moves) one centimeter tall - 200 times shorter than a 6 foot 6 inch human. If the robot was four times as dense as a 70 kilogram human, by scaling it would weigh 35 milligrams. If arms and head moved with the same size-relative speeds, they would weigh 2 million times less, and move 200 times more slowly, an energy scaling of 12.5 parts per trillion. So if a human on average dissipates 8 watts for movement, the tiny robot would dissipate 100 picowatts.
A fast swinging robotic arm, to scale, would move 200 times more slowly than a human arm - centimeters per second - but might weigh a fraction of a gram, and produce 200 nanonewtons of force, which could be done with electrostatic forces. 200 nanonewtons could accelerate a 100 milligram mass to 1 cm/sec speeds in 5 seconds - which scales up to 200 kg at 2 m/s (440 pounds at 4.5 mph ) at human scale.
Some things scale poorly, like eyes - if the robot's eyes produced 1000 by 1000 pixels, with an imaging chip density of 1 pixel per micron, the imager would be a millimeter across, much larger than a scaled retina. The lens would be even larger, and require a high ambient light level to gather enough light to operate quickly.
The zero gee vacuum environment of the "room" will have many cameras, vibration sensors, etc. built into it, so the image seen by the operator on Earth can be a high definition synthesis using all the information available. Our robot does not need to look through a microscope - the microscope image can be transmitted directly to the ground. Larger objects can be moved by larger arms. In many cases, the only function of "rooms" is to contain loose objects, not air, so they can be large vacuum balloons many meters across. Scaled, a room with the volume of Kennedy Space Center's Vertical Assembly building would be 0.5 cubic meters at tiny scale, a sphere a meter in diameter, and if it had aluminum walls 20 micrometers thick it would weigh 160 grams.
This station could house a thousand 35 milligram robots, 35 grams, with another 65 grams of tiny tools and repair parts for them. It would have large, ultrathin solar panel "wings" to provide power. Assume the whole assembly weighs 10 kilograms, and has a sun-facing area of 50 square meters, collecting (and dissipating) 10 kilowatts in full sunlight. Most of that power will drive distributed computation, and large area sensors. In microgravity, the little robots could build gossamer structured instrumentation with wire-thin struts.
The station will have the same area-to-mass ratio as a Server Sky array floating hear it, which will provide computation and communication back to scientists on earth. Four thousand scientists and technicians, working in shifts, could operate hundreds of experiments and manufacturing projects on the station.
This station would be ideal for rebuilding thinsats. Server sky thinsats could be remanufactured by automated assembly lines, with micromachinery maintained by the small telepresence robots.
Many experiments would be too hazardous to operate on earth - microbiology experiments with pathogenic bacteria, for example. A spherical "test tube" holding a thousand 2 micrometer diameter bacteria at 0.1% water concentration would be 0.4 millimeters across, appearing to be about the size of a softball to the tiny robots. The bacteria could be observed with small, scalable devices like scanning tunnelling microscopes, which are intrinsically micrometers in size - our small robots could easily handle the STM components, and move the bacteria through vacuum with electrostatic "tweezers".
Lab rats would be way too big to house and feed, and even high-metabolism shrews would require far too much food and air. The smallest mammal, the two gram Kitti's hog-nosed bat, feeds on the small insects it gathers in less than an hour per day, and might have a suitable metabolism for growth in a closed cycle space environment. The insects themselves might make better experimental animals for studies that do not require a mammal.
These experiments are inspired by Michael Turner's exo vivarium ideas.
Construction and Launch
The 10 kilogram space station would cost a few hundred thousand dollars to launch and deploy, and high automation and much design to build. The materials would be cheap. This seems ideal for a country like India, with tiny budgets but with many scientists and technicians who will work for low wages. After development, most of the manufacturing costs will be for the communication systems and haptic suits for the researchers on the ground - but researchers in other countries will "pay to play", so this will represent revenue, not cost, to the country that deploys it.