Firstly, a single oscillator will tend to produce rather a subjectively 'thin' and static sound. This isn't always the case (as later stages, such as the filter, or a separate chorus stage, can add warmth and movement), and it isn't always a bad thing - but of course if you want to have a one-oscillator sound on a synth that has more oscillators, you can always decline to use more than one!
We're often told that musical sounds are periodic, and that they have overtones that are integer multiples of the fundamental frequency. This often is true of synthesizer oscillators. However, in nature, this is usually only approximately true - a vibrating string will often have higher overtones whose frequencies are a little higher than those multiples, for example. Many acoustic musical sounds also have multiple oscillating elements at slight pitch offsets - the string section is a classic example. Being able to play multiple oscillators, slightly detuned, is therefore useful for producing these subjectively more natural sounds.
You can also take this detuning further and have oscillators tuned in musical intervals (e.g. a fifth apart.) You could also use mix different shapes of oscillator to produce spectra that couldn't be produced by a single simple oscillator.
Possibly the most interesting uses of multiple audio-rate oscillators is to use one to modulate another at an audio rate - in techniques such as PWM, FM, and oscillator sync. These techniques also do a good job of producing more interesting and animated (though not always very natural!) spectra.
I've assumed here from the picture that you're talking about multiple audio-rate oscillators, that would normally be described as 'VCOs' - you'll probably be aware that synths usually have one or more low frequency oscillators (LFOs) that can also be used for modulation purposes.